Oracle History #5

New Features for
Administrators
(Oracle8i)

April, 2001

CloudBox

Jang Kyung Sang

Oracle8i New Features http://www.ggola.com/

Oracle8i New Features for Administrators

Course

1. Java in the database

2. Optimizer and Query Improvements

3. Summary Management

4. Indexes and Index-Organized Tables

5. Basic Partition

6. Partition Maintenance

7. Composite Partitioning

8. Added New Features for partition

9. Universal Installer Migration and Upgrade

10. Tablespace Management

11. Database Resource Manager

12. Manageability Features

13. Availability and Recoverability

14. Features of Net8

15. SQL*Plus, PL/SQL and National Language Support

16. Database Security

17. Partial addition for Oracle8i

jkspark@hanafos.com -2-


1. Java in the database

What does Oracle8i support for java?

.

Open, portable, productive, internet computing language

Oracle8i enterprise class java server
- Java Virtual Machine integrated with Database
- use java anywhere PL/SQL used
- productive programming tools
- industry standard components

Oracle8i java application server platform
application java solution
.

Supported three products for building java database application

- standards-compliant JDBC drivers to access Oracle from JAVA
applications

- precompiler(SQLJ) for embedding SQL in java applications

- Oracle JDeveloper (JDBC, SQLJ, complete development
environment)

CF. Oracle8i JVM(Java
virtual machine) . EJB(Enterprise java bean)
transaction server CORBA clients
.

Five major components that integrated JVM with Oracle8i

- Oracle JVM Javasoft JDK specification Oracle
architecture optimizing .

- Embedded JDBC Driver JVM server
locally SQL and PL/SQL access .

- SQL and PL/SQL Inter-Language Method Services
mechanism JAVA call .
SQL PL/SQL call
C language external procedure ‘call
out’ standard mechanism .
, calling mechanism Oracle8i
SQL and PL/SQL Java call .



- Net8 java program SQL
‘publishes’ SQL Java access
.

- CORBA IIOP(Internet Inter-ORB
Protocol) . RDBMS access second
protocol ( Net8( SQL*Net) first
protocol ) Oracle8i
java CORBA2.0-compliant object request broker
embed CORBA
RDBMS call in and call out .
CF. java stored program ORB ‘published’
clients RDBMS java program access alternate
mechanism .

Major components of JVM

- SQLJ Translator : JAVA and SQL program
.

- Object Memory Management : allocate and free memory in
standard chunks called object memories.

- Memory manager/Garbage collector : Oracle database
environment manager JVM memory
heaps, allocating and collecting object memories
manage .

- Java class loader : Standard “.class” format and in a
compressed form Oracle VM other systems
java binaries interchange . (both export
and import)

- Bytecode compiler : “.class” java binary standard java
program translate .

- Interpretation/run-time : execute standard java binaries.

- Native compilation : C language java program
execution performance .

- Library manager : java program manage .

- Standard library : java API specification(java.lang, java.io,
java.math, and java.util) standard libraries Oracle
Java VM support .

CF. The java interpreters and compilers are based on the JDK 1.1.6
standards



Using JAVA

Integrated Java, SQL, and PL/SQL

- Integration : Java SQL and PL/SQL call ?
SQL and PL/SQL Java call ?

Java call SQL and PL/SQL

- Java standard JDBC interface database SQL
and PL/SQL CALL .

- Java VM OCI thin JDBC drivers interface
standard JDBC driver embed .

- , outside database JDBC
java application code
Oracle Java VM .

- architecture outside or within RDBMS
java and SQL same interface
application partitioning flexibility and efficiency
.

SQL and PL/SQL call java

- SQL and PL/SQL java call SQL
PL/SQL call mechanism .

- Oracle8i java stored procedure PL/SQL
automatic dependency-checking mechanisms
java call .

- Oracle8i new DDL extensions “create
procedure as java” java stored procedure,
function, and triggers .

- Oracle8i PL/SQL namespace java
stored procedure any PL/SQL
.



What is JDBC?

JDBC(Java Database Connectivity)

- JDBC java RDB access
standard set of java classes .

- JDBC javasoft SQL92 syntax
and types . mechanisms java
client database automatic data streaming .

- JDBC Oracle performance improvement
array interface vendor-specific
extensions .

JDBC drivers

- JDBC Thin driver : web browser applets
. (means thin client) driver RDBMS directly
communicating java sockets
client installation .

- JDBC OCI : driver Oracle Call Interface
Net8 RDBMS engine communicating
client OCI library .

- JDBC KPRB : driver java stored procedure
server side OCI Oracle KPRB library
.

- drivers the same syntax and APIs and oracle
extensions java application very
portable . , connect string minor
modification application 3
.

Features in JDBC Drivers

- 3 JDBC drivers Javasoft JDBC 1.22
specification JDBC 2.0 specification
. , JDBC 2.0 specification client side
XA support .
CF. server side driver XA specification
JDBC 2.0 specification .

- Oracle JDBC drivers standard JDBC driver Oracle
specification extensions important features .
1. access to Oracle7 types (like ROWIDs and REFCURSORs)
2. access to Oracle8i object-relational extensions, REFs, Lobs,
and BFILEs



3. performance enhancement with operation to prefetch rows
and the array interface
4. access to PL/SQL stored procedures with Oracle7 and 8
5. access to java stored procedures with Oracle8i
6. access across all Net8 certified firewalls

SQLJ

SQLJ Oracle Pro*C precompilers
Java application embedded SQL
.

Preprocessor embedded SQL statements SQLJ
Java code(with JDBC call) translate .

SQLJ generated code 100% compliant with Javasoft
JDBC specification .

Improve developer productivity

- smaller, cleaner, simpler code than JDBC

- SQL syntax errors and identification of SQL/Java type
mismatches detection code
quality improvement

- JDeveloper Java tool integration

- Standard language specification, compatible translators

- Oracle Translator standard compliant

- SQLJ applications binary portability
(property) platforms different vendor’s database
portable .

SQLJ supports

- SQLJ JDBC application programs
. (concise more than JDBC. easier to write
or maintain)

- 3 SQLJ .
1. JDBC (character string,
prepare and execute statements, and bind and define
variables) directly embed SQL statements
.
2. SQL statements Java bind variables and expressions
.



3. application simple default connection
and other syntactic elements features .

- During the translation phase, SQLJ detects SQL semantic
errors and Java and SQL type mismatches.

- SQLJ provides strong typed queries through the use of SQLJ
Iterators.(first-class java types for cursors)

- SQLJ SQL precompilation techniques run-
time execution more efficient . precompiling
Oracle SQLJ executions .

Code example JDBC VS SQLJ

- JDBC
Java.sql.CallableStatement stmt;
Connection conn;
ResultSet results;

Conn = DriverManager.getConnection 9”jdbc:default”);
stmt = conn.prepareStatement
(“select ename from emp where sal > ? and deptno = ?”);
stmt.setInteger(1, salparm);
stmt.setInteger(2, deptnoparm);
results = stmt.executeQuery();

- SQLJ
ResultSet results ;

#sql results =
{ select ename from emp
where sal > :salparm and deptno = :deptnopark} ;

Using SQLJ with Oracle

- SQLJ JDBC run-time environment deployed
configuration deployed .
CF. SQLJ run-time : Java + JDBC driver

- SQLJ applications deploy JDBC
driver . Oracle or users
JDBC/OCI/thin JDBC driver .

- Oracle8i Java VM embedded SQLJ translator
Oracle SQLJ java
stored procedures and triggers .

- Write program in SQLJ --> SQLJ translator : converts to java +
JDBC --> run the program



Java program

Java stored programs
- open alternative complementing PL/SQL

- java is fully functional. PL/SQL Java
.

- PL/SQL SQL-intensive SQLJ java stored
procedure .

- Java stored procedures 3 run-time contexts
.
1. stored functions and procedures (java program SQL
)
2. all triggers java .
3. java object-relational methods .

Developing

1. write java stored programs

2. Java VM load java code. Java classes
load .
- SQL*Plus “create java ddl”command
- standard Oracle utility “LOADJAVA”

3. publish the java procedure to SQL.
- expose the top-level java entrypoint
- map parameters and SQL and PL/SQL datatypes
- set up appropriate user privileges

4. SQL or PL/SQL call java program
CF. Oracle PL/SQL call SQL or
PL/SQL java program call syntax
.

Example (java installation . Refer to page 11)

create or replace java source named "NewJava" as public class JavaMaster {
static public String Msg(String guys) { return "Good morning " + guys ; } }
/
create or replace function JavaFunc (str varchar2) return varchar2 as
language java name 'JavaMaster.Msg (java.lang.String) return java.lang.String' ;
/
SQL> select JavaFunc('girls') from dual ;
JAVAFUNC('GIRLS')
--------------------------------------------------------------------------------
Good morning girls



Components

Develop server-side business logic as distributed components
- components improve developer productivity in following
ways.

1. reuse server-side business logic

2. improve application design as reusable modules

3. simplify deployment in a multitier environment

Support java component programming models

- Oracle8i traditional CORBA servers Java
implemented . , ORB Java VM java
applications .

- EJB(the industry standard enterprise javabeans)
component model database java
.
Simple declarative transaction commit or rollback
transaction operations .

CF. Java or EJBs implemented CORBA servers
Oracle8i MTS .

Server-side ORB (on MTS)

- IIOP listener : Oracle8i standard CORBA/IIOP binding
protocol extended .(Net8)

- IIOP dispatcher : Oracle8i IIOP-based method
invocations extened .

- Oracle8i integrates a CORBA 2.0 ORB.

- MTS platforms : Java VM MTS .
MTS .

Enterprise JavaBeans (EJBs)

- EJBs is a server-oriented component model for Java.
Oracle8i EJBs transaction server platforms EJB
1.0 specification .

- Significant advantages using EJBs
1. developers advanced system programming
skills .
2. Pure Java definition D/COM, CORBA

jkspark@hanafos.com - 10 -


dependency .
3. platform vendors EJB
.

Java Configuration

Java VM classes installation

- Java VM manually initialize script
.

- $ORACLE_HOME/javavm/install/initjvm.sql
(user by “SYS”)
CF. test 1 .

- script 4000 java classes
initial package .

- DBMS_JAVA package
only used internally by Oracle .

- Needed initial parameter values for executing initjvm.sql :
shared_pool_size 50M
java_pool_size 50M
rollback segment 100M

Java-related initial parameter

- shared_pool_size : static java variables UGA
space

- java_pool_size : shared java objects
space java class loading .
large java application system 50M
java stored procedure 20M
.

- java_soft_sessionspace_limit : size user’s
session java memory usage RDBMS trace file
. default 1M.

- java_max_sessionspace_size : size out-of-
memory failure session kill . Default 4G
invisible java
programs memory usage
.

CF. java-related memory application system java
setting .



Internet File System

iFS Windows OS directories
database access .

Oracle8i Oracle8i Java VM Java
application .

Oracle iFS several different protocols access .

- SMB : MS windows95, 98, NT protocol

- HTTP : Web browser and network computer

- FTP : FTP clients command line

- SMTP, IMAP4, POP3 : e-mail protocol

Data integration

- iFS application developers 3
APIs , Java, CORBA, and PL/SQL .

- iFS document system document
automatically decompose(parse)
document recompose(render) .

- XML(Extended markup language) .

CF. iFS application
.



2. Optimizer and Query Improvements

Stored Outline

Optimizer plan stability

- Optimizer feature Oracle8i
well-tuned applications SQL hints
(good access path) SQL
.

- Plan Equivalence : plan
1. SQL Statement text must match
2. Plans are maintained through :
- new oracle versions
- new statistics on objects
- init.ora parameter changes
- database reorganization
- schema changes

CF. table index stored outline index
.

Creating stored outlines

- Outline oracle automatic or SQL
creation rule-based or cost-based
optimizers .
oracle outline cost-based optimize
.

- Session init.ora parameter create_stored_outlines
parameter true default category
category name category stored
outlines .

- Test initial parameter database
restart error parameter
create_stored_outlines

SVRMGRL> show parameter
SQL> select name from v$parameter
. Oracle8.1.5 Reference
parameter session level
test .

CF. system user “create any outline” privilege outline
user grant .



- Session
category name default
category name .

SQL> alter session set create_stored_outlines = true ;
Session altered.
SQL> select d.dname, e.ename from emp e, dept d
2 where e.deptno = d.deptno ;
DNAME ENAME
-------------- ----------
RESEARCH12 SMITH
SALES12 ALLEN
..............
ACCOUNTING12 MILLER
14 rows selected.………..
SQL> alter session set create_stored_outlines = false ;
SQL> select * from user_outlines ;
NAME CATEGORY USED TIMESTA VERSION
SQL_TEXT
----------------------------------------------------------------------------------------------------------------
SYS_OUTLINE_0107311031050000 DEFAULT UNUSED 31Jul01 8.1.6.2.0
select d.dname, e.ename from emp e, dept d where e.deptno = d.deptno

SQL> alter session set create_stored_outlines = cat_emp ;
Session altered.
SQL> select e.ename, d.loc from emp e, dept d where e.deptno = d.deptno ;
ENAME LOC
-------------- ----------
SMITH DALLAS189
ALLEN CHICAGO189
..............
MILLER NEW YORK189
14 rows selected.………..
SQL_TEXT
-------------------------------------------------------------------------------- -------------------------------
SYS_OUTLINE_0107311026420001 CAT_EMP UNUSED 31Jul01 8.1.6.2.0
select e.ename, d.loc from emp e, dept d where e.deptno = d.deptno
select * from user_outlines
SQL> alter session set create_stored_outlines = false ;

- category DEFAULT
category
. outline name
system default
.

- Outline parameter false select * from outlines
stored outline .

- outline category shared_pool pinning
execution plan .



- stored outline .

SQL> create or replace outline emp_join for category cat_emp
on select /*+ FULL(DEPT) */ d.dname, e.ename from emp e, dept d
where e.deptno = d.deptno order by 1, 2 ;
Outline created.
SQL_TEXT
----------------------------------------------------------------------------------------------------------------
select e.ename, d.loc from emp e, dept d where e.deptno = d.deptno
select * from user_outlines
EMP_JOIN CAT_EMP UNUSED 31Jul01 8.1.6.2.0
select /*+ FULL(DEPT) */ d.dname, e.ename from emp e, dept d
where e.deptno = d.deptno order by 1, 2

- outline name outline sys
name . category
.

category .

Using stored outlines

- outline command .

SQL> alter session set use_stored_outlines = true or category_name ;

- create category name true
outline .

- Category category
default category text match .
, category match outline
default category outline outline
.

- To determine a SQL statement’s execution plan
1. shared pool matching text and outline category
2. 1 data dictionary outline query
3. 2 SQL outline execution
plan .
4. 2 normal methods(non-outline) .

- , outline hints outline
user_outlines user_outline_hints
view
.



Maintaining stored outlines
- Package ‘OUTLN_PKG’
1. drop_unused : drop unused outline
2. drop_by_cat : drops categories of outlines
3. update_by_cat : rename categories

- ALTER OUTLINE outline_name command
1. rename to new_outline_name ; (outline name )
2. rebuild ; (outline execution path )
3. change category to new_category_name ; (category )

Database creation automatically - Stored outlines in tables in schema OUTLN
schema OUTLN 1. OL$ : outline name, category, timestamp, statement text
tables outlines . 2. OL$HINTS : hints for the outlines in OL$
(system tablespace )
- Data dictionary view
1. DBA/ALL/USER_OUTLINES
2. DBA/ALL/USER_OUTLINE_HINTS

DBMS_STATS

This package enables you to generate and manage statistics for
cost-based optimization. The statistics can reside in the data
dictionary or in a table created in the user’s schema. Only
statistics stored in the dictionary itself will have an impact on
the cost-based optimizer. (but cannot gather cluster statistics)

Generating statistics
- Gathering sub-procedure
1. gather_index_stats : index
2. gather_table_stats : table, column and index
3. gather_schema_stats : all objects in a schema
4. gather_database_stats : all objects in a database

- usage for table statistic
DBMS_STATS.GATHER_TABLE_STATS (
schema, table, partition, sample size(%), block sample?(or row),
columns, parallelism degree, table and all partitions,
cascade to indexes) ;

- How can you gather the statistics ?
1. statistics serially or in parallel index
statistics only serially
2. random sampling or computed
3. block sampling faster than row sampling
4. column lists
5. table index statistics
6. statistics tables can hold statistics to enable you to
experiment with different sets of statistics.



Moving statistics

scott emp table export
file dump database import statistic
import source database target database
.

1. table statistics .
usage : dbms_stats.
gather_table_stats(schema, table, partition, sample size(%), block
sample?(or row), columns, parallelism degree, table and all partitions,
cascade to indexes)
meaning : table statistic . (index statistics
cascade to indexes TRUE )
example :
SQL> exec dbms_stats.gather_table_stats( 'SCOTT', 'EMP',
NULL, 20, FALSE, 'FOR ALL COLUMNS', 4, 'DEFAULT',
TRUE) ;

2. Statistics table schema .
usage : dbms_stats.
create_stat_table(‘SCOTT’, ’stat_tab_name’, ’tablespace_name’)
meaning : data dictionary statistics export
table .
example :
SQL> exec dbms_stats.create_stat_table('SCOTT',
'STAT_SCOTT', 'USERS') ;

3. Statistics data table copy .
usage : dbms_stats.
export_table_stats(‘SCOTT’, ‘EMP’, NULL(partition ),
‘stat_tab_name’, ‘stat_id’(ID ), TRUE(index stat) ) ;
Source Database meaning : statistics user’s table export .
table export backup or copy .
example :
SQL> exec dbms_stats.export_table_stats('SCOTT', 'EMP',
NULL, 'STAT_SCOTT', 'Stat 20010731', TRUE) ;
File dump exp/imp
4. database table file dump
OR export table import database
Database link copy link copy table ,
stat_scott table database
data data dictionary
.
usage : dbms_stat.
Target Database import_table_stat (export_table_stas argument)
meaning : schema table data dictionary
.
example :
SQL> exec dbms_stats.import_table_stats('SCOTT', 'EMP',
NULL, 'STAT_SCOTT', 'STAT 20010731', TRUE) ;



5. user schema table
drop .
usage : dbms_stats.
drop_stat_table(‘SCOTT’, ‘stat_tab_name’)
meaning : table drop .
example :
SQL> exec dbms_stats.drop_stat_table ('SCOTT',
'STAT_SCOTT') ;

Top-N SQL

Now Oracle8i supports ordering of views and in-line views.

Sorts only the required number of rows

- Oracle8i in-line view
order by .

- Oracle rownum < 3
2 value memory rows
2 value .

- Example : order by
SQL> select * from
(select empno, deptno, hiredate from emp
order by hiredate desc)
where rownum < 3 ;

EMPNO DEPTNO HIREDAT
---------- ---------- -------
7876 20 12Jan83
7788 20 09Dec82

SQL> select * from
(select empno, deptno, hiredate from emp)
where rownum < 3 ;

EMPNO DEPTNO HIREDAT
---------- ---------- -------
7369 20 17Dec80
7499 30 20Feb81

- Query (searching for the top few values) run considerably
faster in Oracle8i.



ROLLUP and CUBE

Aggregate operators

group by rollup and cube
crosstab report .

1 2 3 Total
1 2 3 6
1 2 3 6
Total 2 4 6 12

Rollup operation

SQL> select , , sum(year) from emp
group by rollup( , );

- group by total grand total
.

sum(year)

1 1
2 2
3 3
6
1 1
………………..
………………..
6
………………..
12

- function data result rows n types( ) m
status( ) ROLLUP group by (n X
(m+1)) + 1 = ? rows .

- (2 X (3+1)) + 1 = 9 rows .



Cube operation

SQL> select , , sum(year) from emp
group by cube( , )
order by , ;
sum(year)

1 1
2 2
ROLLUP 3 3
6
1 1
2 2
+ 3 3
6
CUBE 1 2
2 4
3 6
GRAND 12

- function data result rows n types( ) m
status( ) CUBE group by ((n+1) X
(m+1)) + 1 = ? rows
.

- .
, (type + 1) X (status + 1) .
(2 + 1) X (3 + 1) = 12 .

Grouping function

Rollup and cube total value null
. data value null
null .
new function grouping .

- grouping(column_name) decode 0
value null 1 rollup or cube
null .

- , .
decode(grouping(column),1,’null of total’,’value’)



TRIM

New function trim ltrim rtrim .

ANSI .

Using function
SQL> select last_name, trim(leading ‘X’ from last_name)
as trimmed
from emp where last_name like ‘X%’ ;

- TRIM(leading ‘X’ from column) : LTRIM

- TRIM(trailing ‘X’ from column) : RTRIM

- TRIM(both ‘X’ from column) : L + R (default)

- TRIM( ‘X’ from column) : both .

- TRIM(last_name) : character
space trim characters .

CF. trim character .
trim character only one character .

New and Obsolete

Obsolete parameter :
sort_direct_writes, sort_write_buffers,
sort_write_buffer_size, sort_read_fac

New parameter :
sort_multiblock_read_count
this parameter define the number of blocks read from
temporary segments during sorts.
(db_multiblock_read_count )



3. Summary Management

Materialized Views

In general, this will be used for Data Warehouse, Distributed
Computing, Mobile Computing

Summary and aggregate data

- View view base table data view
, summary and aggregate data(pre-computed)
data segment view .

- , expensive and complex joins view .

Create materialized view day_of_booking as
Select bkg_no, sum(amt) as total_amount
From booking
Where bkg_dt = sysdate
Group by bkg_no ;

- M-View SQL base
table M-View
.( query rewrite )
CF. The process of modifying a query to use the view rather
than the base table is called a query rewrite.

- Create materialized view is synonym for create snapshot.

- M-View
SQL> create materialized view view_name build deferred
refresh …. as select …..
“build” clause M-View
. Build clause “immediate or
deferred” option .
1. immediate
2. deferred refresh .
CF. build clause default “build
immediate” .

CF. remote tables .

CF. M-View order by clause
creation
refresh . , order by
clause M-View query definition .
query performance order by clause
column M-View select list column
index recommend .



Materialized view refresh option

Base table M-View base table sync
. ‘Refresh’ .

- Complete : Base table
M-View
M-View query
data .

- Fast : last refresh
1. M-View log
2.Direct loader log (direct loads new rows
rowid M-Views refresh)

- Force : .
complete fast .
(This is default refresh type.)

- Never : Refresh .

CF. M-View system user user grant
‘create materialized view’ privilege .

Refresh mode : Refresh .

- ON DEMAND
Manually DBMS_MVIEW package
procedure .
1 …REFRESH(‘BOOKING’,PARALLELISM => 3);
>> specified M-View
2 …REFRESH_DEPENDENT(VN_FAIL_CNT, ‘EMP’) ;
>> M-Views on based 1 base tables
(number_of_failures out variable )
3 …REFRESH_ALL_MVIEWES ;
>> all M-Views

- Automatic Refresh
1. ON COMMIT ( synchronous) : base table
commit .
2. START WITH-NEXT(asynchronous) : specified time
supported complete mode only. (must initiate job
processes)

CF. , ON COMMIT M-View join or aggregate
views Oracle8.1.5 manual
only join view single table aggregate
.( Nested M-View
join and aggregate view fast on commit
)



Materialized view types (UNION set operation )

- joins and aggregates
type M-View fast refresh
new data direct path method add fast
refresh . The where clause can contain inner
equi-joins only.(no outer joins)
1. group by column select list .
2. function : sum, count(x), count(*), count(distinct x), avg,
variance, stddev, min, max

- single table aggregate
M-View log fast refresh count
sum count count or sum
group function count sum
. count(*) where
clause .
1. function : count, sum, avg, variance, stddev

- only joins : fast refresh
1. each detail tables M-View log
2. select list detail table rowid .
3. outer join inner table column
unique constraint .
CF. M-View log rowid column
speed up refresh M-View rowid
column index recommend .
CF. detail tables rowid tables
tables DML fast refresh .

Test M-View

- Test Oracle8.1.6.2 M-View
. Oracle Manual

.

- , Oracle version
Oracle9i fixed Test
.

- Refresh complete .
dbms_mview package refresh

M-View
.

- Refresh fast on commit on demand
M-View log Test
Query Rewrite Usage
.



Example
- complete on demand (joins and aggregates)
SQL> sho user
USER is "SYSTEM"

SQL> grant create materialized view to scott ;
Grant succeeded.

SQL> conn scott/tiger
Connected.

SQL> set timing on
SQL> set serveroutput on

SQL> create materialized view mv_emp_sales refresh complete on demand
2 as select e.empno, d.deptno, sum(sal) from emp e, dept d
3 where e.deptno = d.deptno group by e.empno, d.deptno ;
Materialized view created.
Elapsed: 00:00:46.95

SQL> select * from mv_emp_sales ;
EMPNO DEPTNO SUM(SAL)
---------- ---------- ----------
7369 20 52428800
....................
9022 95 6553600
21 rows selected.

EMPNO SAL update SQL> update emp set sal = 10 where empno = 9022;
256 rows updated.
.
SQL> commit ; MV creation
Commit complete. refresh
.
---------- ---------- ----------
7369 20 52428800
....................
9022 95 6553600
21 rows selected.
Refresh procedure
SQL> exec dbms_mview.refresh('MV_EMP_SALES') ;
PL/SQL procedure successfully completed.
. Elapsed: 00:00:48.04


---------- ---------- ----------
7369 20 52428800
....................
9022 95 655360
21 rows selected.

CF. M-View performance M-View
partitioning parallel operation
.



SQL> update emp set sal = 100 where empno = 9022 ;
256 rows updated.

SQL> commit ;
Commit complete.

refresh_dependent procedure SQL> declare
2 vn_fail_cnt BINARY_INTEGER;
3 begin
4 dbms_mview.refresh_dependent(vn_fail_cnt, 'EMP');
. 5 dbms_output.put_line(vn_fail_cnt);
6 end;
7 /
0

Elapsed: 00:00:47.74


---------- ---------- ----------
7369 20 52428800
....................
9022 95 6553600
21 rows selected.

- fast on commit (single table aggregate)
SQL> create materialized view log on emp
2 with rowid(empno, ename, job, mgr, hiredate, sal, comm, deptno)
3 including new values ;
Materialized view log created.

SQL> create materialized view mv_emp_sales refresh fast on commit
2 as select count(*), empno, sum(sal), count(sal)
3 from emp
log M- 4 group by empno ;
View . base Materialized view created.
table SQL> select * from mv_emp_sales ;
refresh COUNT(*) EMPNO SUM(SAL) COUNT(SAL)
commit refresh ---------- ---------- ---------- ----------
data 256 7369 204800 256
...........................
. 256 9022 25600 256
21 rows selected.

256 rows updated.

SQL> commit;
Commit complete.

COUNT(*) EMPNO SUM(SAL) COUNT(SAL)
---------- ---------- ---------- ----------
256 7369 204800 256
...........................
256 9022 2560 256
21 rows selected.



- fast on demand (single table aggregate)

SQL> drop materialized view log on emp ;
Materialized view log dropped.

SQL> drop materialized view mv_emp_sales ;
Materialized view dropped.

SQL> create materialized view log on emp
2 with rowid(empno, ename, job, mgr, hiredate, sal, comm, deptno)
3 including new values ;
Materialized view log created.

SQL> create materialized view mv_emp_sales refresh fast on demand
2 as select count(*), empno, sum(sal), count(sal)
log 3 from emp
MV . base table 4 group by empno ;
Materialized view created.

refresh procedure SQL> select * from mv_emp_sales ;
MV
COUNT(*) EMPNO SUM(SAL) COUNT(SAL)
---------- ---------- ---------- ----------
data . 256 7369 204800 256
.........................
256 9022 2560 256
21 rows selected.

256 rows updated.

SQL> commit;
Commit complete.

SQL> exec dbms_mview.refresh('MV_EMP_SALES') ;


Test join and COUNT(*) EMPNO SUM(SAL) COUNT(SAL)
---------- ---------- ---------- ----------
aggregate M-View 256 7369 204800 256
M-View log .........................
error fast refresh 256 9022 256000 256
21 rows selected.
refresh
data CF. M-View table
. prebuilt table M-View .
query rewrite . (start with-next refresh
M-View query rewrite )

CF. Nested Materialized Views : M-View fact(/detail/base)
table M-View user request
Nested Materialized View .
join only M-View base
table single-table aggregate M-View refresh
fast on commit join and aggregate M-View
performance . ,
space .



Query Rewrite

M-View base table
performance .
Oracle8i M-View statement ‘Query
Rewrite’ mechanism M-View
query query .

Query Rewrite Attributes

- M-View index transparent to
applications .

- Base table privileges M-View
M-View enable,
disable M-View control .

- Initial parameter : QUERY_REWRITE_ENABLED=TRUE
”QUERY REWRITE” privilege M-View enable .
DBMS_OLAP package has options to use M-Views

CF. Query Rewrite : grant query rewrite to username ;
“global query rewrite” for outside schema
“explicit select privilege” for M-View using different schema

Usage :
create materialized view mview_name
tablespace ts_name --> skip user default tablespace
parallel (degree 4) --> option
build immediate/deferred --> skip default immediate
refresh complete/fast/force/never --> skip default force
on demand/on commit/start with .… next ….
enable/disable query rewrite --> query rewrite
as select count(*), count(sal), sum(sal) from emp; --> query
restriction : join equip only aggregate
operator outermost part . (AVG(AVG(x))
) sequence, user, sys owner objects,
subquery order by, remote table, pl/sql function, having,
connect by, inline views, nested subquery, sysdate, rownum,
bind variable, raw or long raw .

case 1) select count(*), count(sal), sum(sal) from emp ; ==>
Oracle M-View
mview_name execution plan
performance .

case 2) select /*+ NOREWRITE */ count(),count(sal),sum(sal) from
emp ; ==> Case 1 norewrite hint
M-View base table execution
plan full table scan .



Example
- Creation materialized view for query rewrite
SQL> create materialized view mv_emp_sales tablespace users
2 parallel(degree 4) build immediate
start and next time 10 3 refresh complete start with sysdate+1/6/24 next sysdate+1/6/24
query 4 enable query rewrite as select e.empno, d.deptno, sum(sal) from emp e, dept d
rewrite enable 5 where e.deptno = d.deptno group by e.empno, d.deptno ;
MV Materialized view created.
Elapsed: 00:01:02.80
user_jobs
last date, next date 10 SQL> select job, what, last_date, next_date from user_jobs ;
JOB WHAT
---- -----------------------------------------------
.
243 dbms_refresh.refresh('"SCOTT"."MV_EMP_SALES"');
LAST_DATE NEXT_DATE
----------------- -----------------
20010802 14:35:31 20010802 14:45:31
SQL> select sysdate from dual ;
SYSDATE
-----------------
20010802 14:36:41
14 36 ---------- ---------- ----------
empno 9022 sal 10 7369 20 52428800
................
update . 9022 95 6553600
MV query 21 rows selected.

. user_jobs
256 rows updated.
next date SQL> commit ;
. Commit complete.

---------- ---------- ----------
7369 20 52428800
................
9022 95 6553600
38
21 rows selected.
refresh SQL> select sysdate from dual ;
. 45 SYSDATE
refresh -----------------
MV 20010802 14:38:20
( 1 )
46 40 query SQL> select sysdate from dual ;
SYSDATE
data
-----------------
. 20010802 14:46:41
, refresh
MV query SQL> select * from mv_emp_sales ;
“no rows selected” return EMPNO DEPTNO SUM(SAL)
. ---------- ---------- ----------
7369 20 52428800
................
9022 95 655360
21 rows selected.



- Query rewrite
SQL> select e.empno, d.deptno, sum(sal) from emp e, dept d
---------- ---------- ----------
MV SQL 7369 20 52428800
................
9022 95 655360
. ,
21 rows selected.
query rewrite .
Elapsed: 00:00:00.08
hint
query rewrite SQL> select /*+ NOREWRITE */
2 e.empno, d.deptno, sum(sal) from emp e, dept d
. 3 where e.deptno = d.deptno group by e.empno, d.deptno ;
---------- ---------- ----------
7369 20 52428800
................
9022 95 655360
21 rows selected.
Elapsed: 00:00:49.50
SQL> explain plan for select e.empno, d.deptno, sum(sal) from emp e, dept d
Explained.
SQL> select substr(lpad(' ',2*(level-1))||operation||' '||options
2 ||' '||object_name,1,79) "Plan View"
3 from plan_table
4 start with id=0
5 connect by prior id = parent_id
6 order by id;
Plan View
-------------------------------------------------------------------------------
SELECT STATEMENT
TABLE ACCESS FULL MV_EMP_SALES
SQL> delete plan_table;
plan
2 rows deleted.
query rewrite SQL> commit;
MV execution Commit complete.
plan hint SQL> explain plan for select /*+ NOREWRITE */
2 e.empno, d.deptno, sum(sal) from emp e, dept d
execution plan 3 where e.deptno = d.deptno group by e.empno, d.deptno ;
. Explained.
3 from plan_table
4 start with id=0
6 order by id;
Plan View
-------------------------------------------------------------------------------
SELECT STATEMENT
SORT GROUP BY
NESTED LOOPS
TABLE ACCESS FULL EMP
INDEX RANGE SCAN XPKDEPT



Control Rewrite

- optimizer mode cost-based only . (rule base )

- QUERY_REWRITE_ENABLED : FALSE setting query
rewrite . dynamically instance
level(initial parameter) or session level(alter session
command) .

- QUERY_REWRITE_INTEGRITY : this parameter reset
dynamically for instance and session level.
1. ENFORCED(default) : consistency
query rewrite . (updated M-view and validated
constraints only)
2. TRUSTED : All updated M-Views and dimensions and
constraints with RELY flag are used for query rewrites.
(RELY flag constraint new features
invalid constraint valid
)
3. STALE_TOLERATED : last DML and relationships(are
declared) refresh all M-Views
rewrite .(data M-View validate )

CF. register M-View
QUERY_REWRITE_INTEGRITY parameter value
.

Support utility & System privileges

- SQL*Loader : Rowid range M-View
reference table direct path loading
HWMK data incremental refresh
.

- Export/Import :
1. M-View definition export & import (full mode or user mode)
2. MVDATA=Y export & import data in M-Views
MVDATA=N table ROWS=N .
( , M-View data segment
object )
3. MVBUILDDEFERRED=Y
export all M-Views definition and created for deferred
refresh

- M-Views privileges
. create materialized view
. create/drop/alter any materialized view
. query rewrite (enable and disable of own schema)
. global query rewrite (enable and disable of in the grantee’s
schema for query rewrite)



Registration of an Existing Materialized View

- Data Warehouse system ordinary user
tables summary and aggregate data tables
. tables table
build cost
. , data .

- table Query Rewrite
incremental refresh tables
M-View .

- M-View base table
M-View .( name table name )
SQL> create materialized view ….name on prebuilt table …

- prebuilt table M-View query
rewrite parameter query_rewrite_integrity
“TRUSTED” set .

- QUERY_REWRITE_INTEGRITY : (Oracle 8.1.6 )
enforced : M-View data relationship
query rewrite .
trusted : prebuilt table M-View data and RELY constraint
(17 constraints ) correct trust query
rewrite .
stale_tolerated : M-View valid query rewrite
maximum rewrite capability . , query
rewrite wrong result .
CF. Oracle 8.1.5 enforce/no_enforce/use_stale .

- , prebuilt table M-View drop table
drop .

Data dictionary for M-Views
- ALL_REFRESH_DEPENDENCIES :
dependent or container tables in the current schema

- DBA_MVIEW_AGGREGATES :
grouping functions in an aggregated M-Views

- DBA_MVIEW_ANALYSIS : M-Views information

- DBA_MVIEW_DETAIL_RELATIONS :
the named detail relations (base(fact) tables information)

- DBA_MVIEW_JOINS : information about the joined columns
of base tables

- DBA_MVIEW_KEYS : detail columns information about the
relations of DBA_MVIEW_DETAIL_RELATIONS



Dimension

Additional query rewrites are possible if summaries are
created at the lowest level of aggregation, supported with
dimensions.

What is a Dimension ?
- data dictionary structures that define hierarchical relationship
between pairs of columns and column sets. based on existing
columns(M-View segment )

- dimension optional highly recommended
1. enable additional query rewrite possibilities
(without use of constraints, M-Views )
2. hierarchies( ) summary .
3. can be used by OLAP tools ( DBMS_OLAP )

- level dimension . date
month month quarter quarter year
level query
rewrite .
CF. create dimension attribute level
level value column
.( “ ” level
“ ” )

SQL> conn system/manager
Connected.

SQL> grant create dimension to dimuser ;
Grant succeeded.

SQL> conn dimuser/dimuser
Connected.

SQL> create table emp (
2 hiredate date,
3 month varchar2(2),
4 month_name varchar2(10),
5 quarter varchar2(1),
6 year varchar2(4)) ;
Table created.

SQL> Create dimension emp_dim
2 Level hiredate is emp.hiredate
3 Level month is emp.month
4 Level qtr is emp.quarter
5 Level yr is emp.year
6 Hierarchy calendar_rollup (
7 Hiredate child of
8 Month child of
9 Qtr child of yr )
10 Attribute month determines month_name ;
Dimension created.



- 3 multiple hierarchies time
dimension .
SQL> create table sales (
2 sal_date date,
3 month varchar2(2),
4 quarter varchar2(1),
5 year varchar2(4),
6 season varchar2(6),
7 week varchar2(1),
8 dayweek_name varchar2(3),
9 month_name varchar2(3)) ;

Table created.

SQL> create dimension sales_dim
2 level sal_date is sales.sal_date
3 level month is sales.month
4 level quarter is sales.quarter
5 level year is sales.year
6 level season is sales.season
7 level week is sales.week
8 hierarchy cal_rollup (
9 sal_date child of
10 month child of
11 quarter child of
12 year )
13 hierarchy week_rollup (
14 sal_date child of
15 week )
16 hierarchy season_rollup (
17 sal_date child of
18 season )
19 attribute sal_date determines sales.dayweek_name
20 attribute month determines sales.month_name ;

Dimension created.

SQL> select * from user_dimensions ;

OWNER DIMENSION_NAME INVALID REVISION
---------- -------------------- - ----------
DIMUSER EMP_DIM N 1
DIMUSER SALES_DIM N 1

CF. time level, category level, location level
.( , , )



- Based on multiple tables : 2 table
1. Keys and attributes correspond to columns
2. All attributes and keys for one level must belong to one table
3. child columns FK define
.
Create dimension nation_dim
Level goo_cd is goo.goo_cd
Level city_cd is city.city_cd
Level do_cd is dochung.do_cd
Hierarchy nation_analysis (
Goo_cd child of
City_cd child of
Do_cd
Join key goo.city_cd references city.city_cd
Join key city.do_cd references do_cd)
Attribute city_cd determines (city.city_name,
city.city_mayor) ;

Query Rewrite Situations

- Exact match : selected columns and join conditions match

- Aggregate rollup : summary grouping “lower than” query
grouping

- Aggregate to all : more dimensions in summary

- Summary joinback : a column not in summary, but derivable
used by query

CF. exact match dimension
query rewrite .

Data dictionary for Dimensions

- DBA_DIMENSIONS : dimension objects

- DBA_DIM_LEVELS : dimension levels

- DBA_DIM_LEVEL_KEY : column and position of dimension
level

- DBA_DIM_ATTRIBUTES : relationships between a dimension
level and a functionally dependent column

- DBA_DIM_HIERARCHIES : dimension hierarchies

- DBA_DIM_CHILD_OF : hierarchical relationships (a pair of
levels in a dimension)

- DBA_DIM_JOIN_KEY : information between two dimension
tables



DBMS_OLAP

Verifying relationships in a dimension
Satisfy the 1:n relationships along each hierarchy.

DBMS_OLAP.VALIDATE_DIMENSION

- Argument : dimension, owner, check rows, check level

- (‘TIME_DIM’, ’SCOTT’, FALSE, TRUE)
.
1. dimension name ‘TIME_DIM’
2. dimension owner ‘SCOTT’
3. checking all rows
4. check that levels are all non-NULL

- 3 TURE : check rows stored by direct path loads
4 FALSE : no check levels

- It will update MVIEW$_EXECPTIONS with type of
relationship and rowids of violating rows

Evaluate the use of summaries

DBMS_OLAP.EVALUATE_UTILIZATION ;

- uses hypothetical workload statistics
( )

- MVIEW$_EVALUTAIONS .

DBMS_OLAP.EVALUATE_UTILIZATION_W ;

- uses workload statistics collected through Oracle Trace in
WORK$_IDEAL_MVIEW, WORK$_MVIEW_USAGE

- MVIEW$_EVALUTAIONS .

MVIEW$_EVALUTAIONS :

- summary name, storage used by summary in bytes

- frequency of use, rank based on cost-benefit ratio

- cumulative benefit computed by adding the benefit or net
reduction factor

- benefit to cost ratio computed as the cumulative benefit to
storage in bytes



Obtaining summary recommendations

M-Views . (summary advisor)

DBMS_OLAP.RECOMMEND_MV_W

- Argument : table, M-Views bytes
maximum, retain M-View list, retention percent(cumulative
space this percent threshold retain .)

- (‘SALES’,102400000, NULL, 80)
1. table SALES
2. maximum 100M for storing M-view
3. retain M-Views
4. cumulative space 80% .

- MVIEW$_RECOMMENDATIONS
action .

Select recommended_action, mview_name, group_by_columns,
measures_list
From mview$_recommendations ;

RECOMM MVIEW_NAME GROUP_BY MEASURE
RETAIN SALES_SUM
DROP GRP_SUM
RETAIN BRANCH_SUM,
…..
….
CREATE EMPNO,PRO_CD SUM(SALARY)

Estimating storage requirements for a summary or M-Views
DBMS_OLAP.ESTIMATE_SUMMARY_SIZE (space requirement)

- Argument : statement id(in), query(in), estimated rows(out),
estimated bytes(out)

- (‘TRY1’, ‘select * from emp where
dept_id = 7843’, erows, ebytes)
1. statement id for explain : TRY1
2. query emp
3. estimated rows erows return
4. estimated bytes ebytes return

- out argument erows, ebytes
print storage requirements
.

- Query query
string .



4. Indexes and Index-Organized Tables

Bitmap Index

Improvement

- The bitmap is compressed before it is stored in the index.

- compression : value change uncompress and
change bitmap and compress .

- ”alter table table_name minimize records_per_block” this
command block records
(nominimize . This is default)
bitmap index compress .
, empty table IOT nested table
bitmap index
.

- Local Index bitmap index partition
.

Example

- bitmap index and normal index
SQL> sho user
USER is "SCOTT"
SQL> select count(*) from emp ;
COUNT(*)
----------
172032

SQL> create index xakemp_deptno on emp (deptno) ;
Index created.

SQL> select sum(bytes)/1024/1024,sum(blocks) from user_extents
2 where segment_name = 'XAKEMP_DEPTNO' ;
SUM(BYTES)/1024/1024 SUM(BLOCKS)
-------------------- -----------
2.5M 339block 2.6484375 339

index bitmap index SQL> drop index XAKEMP_DEPTNO ;
Index dropped.
storage
SQL> create bitmap index xakemp_deptno on emp (deptno) ;
. bitmap index Index created.
compress
. SQL> select sum(bytes)/1024/1024,sum(blocks) from user_extents
0.3M 45block -------------------- -----------
.3515625 45



SQL> drop index XAKEMP_DEPTNO ;
Index dropped.

SQL> alter table emp minimize records_per_block ;
Table altered.
Minimize command
. SQL> create bitmap index xakemp_deptno on emp (deptno) ;
data Index created.
compress
. SQL> select sum(bytes)/1024/1024,sum(blocks) from user_extents
-------------------- -----------
.3515625 45

New Index

Reverse-Key Indexes (Oracle8 New Features )

- Last leaf block .

- , data 10, 11, 12, 13, 14….
delete leaf block
. (skewed index)
reverse index . alter command
. (01, 11, 21, 31, 41…. )

- cannot support bitmap index and IOT

- “=” index range scan
index full scan .

- Example

SQL> create index xakemp_empno_rvs on emp (empno) reverse ;
Index created.

SQL> explain plan for select /*+ INDEX(EMP) */ distinct empno, ename
from emp
2 where empno > 7800 and empno < 7900 ;
Explained.

3 from plan_table
4 start with id=0
“=” 5 connect by prior id = parent_id
Range Scan 6 order by id;
Index Full Scan
. Plan View
-------------------------------------------------------------------------------
SELECT STATEMENT
SORT UNIQUE
TABLE ACCESS BY INDEX ROWID EMP
INDEX FULL SCAN XAKEMP_EMPNO_RVS



2 rows deleted.
SQL> commit;
Commit complete.

SQL> select /*+ INDEX(EMP) */ distinct empno, ename from emp
EMPNO ENAME
---------- ----------
7839 KING
7844 TURNER
7876 ADAMS

SQL> alter index xakemp_empno_rvs rebuild noreverse ;
Index altered.

SQL> explain plan for select /*+ INDEX(EMP) */ distinct empno, ename
from emp
Explained.

3 from plan_table
4 start with id=0
6 order by id;
rebuild index
Range Scan Plan View
. -------------------------------------------------------------------------------
SELECT STATEMENT
SORT UNIQUE
TABLE ACCESS BY INDEX ROWID EMP
INDEX RANGE SCAN XAKEMP_EMPNO_RVS

2 rows deleted.
SQL> commit;
Commit complete.



Function-Based Indexes

- Index where
dramatic performance
.

- , “ salary – tax “ function based index

select acct_cd from tax_org where (salary – tax) > 100000 ;
data
.

- Enabling Function-Based Indexes
alter session set query_rewrite_enabled = TRUE ;
alter session set query_rewrite_integrity = TRUSTED ;

- Example
SQL> set timing on
SQL> select sum(sal-comm) from emp where (sal-comm) > 3000 ;
SUM(SAL-COMM)
-------------
391577600
Elapsed: 00:00:03.46
Function based index
SQL> create index xakemp_salcnt_fnc on emp ( sal - comm ) ;
Index created.
query rewrite SQL> select sum(sal-comm) from emp where (sal-comm) > 3000 ;
SUM(SAL-COMM)
-------------
. data 391577600
performance Elapsed: 00:00:00.55
SQL> alter session set query_rewrite_enabled=false ;
.
Session altered.
SQL> select sum(sal-comm) from emp where (sal-comm) > 3000 ;
SUM(SAL-COMM)
-------------
391577600
Elapsed: 00:00:03.18

Descending Indexes

- order index
.

- city city index
, deptno
descending index sort
.

- Index : create index city_margin_dsc on emp (city asc, (revenue
index order – cost) desc ) ;
by clause . Query : select city, deptid, ordid, (revenue – cost) margin
from emp
order by city , margin desc ;



- Example

SQL> select ename, sum(sal-comm) from emp where ename like 'K%'
2 and sal-comm > 3000 group by ename
3 order by 1 asc, 2 desc ;

ENAME SUM(SAL-COMM)
---------- -------------
ename sal-comm
KANG 39321600
index KIM 80281600
. , function KING 32768000
based index index
order by Elapsed: 00:00:06.27
index
SQL> create index xakemp_name_salcnt_desc on emp
.
2 (ename asc, (sal - comm) desc) ;
Index created.

2 data order SQL> select ename, sum(sal-comm) from emp where ename like 'K%'
2 and sal-comm > 3000 group by ename
by desc 3 order by 1 asc, 2 desc ;
order by . ,
order by clause 2 ENAME SUM(SAL-COMM)
asc order ---------- -------------
by 1, 2 KANG 39321600
. KIM 80281600
KING 32768000
Elapsed: 00:00:00.94

Rebuild and compute index

Rebuild and Create Index Online

- Oracle8i index rebuild or create online option
DML . lock
time DML
. .

- Parallel rebuild
index 2 space .

- PREPARE : create index structure and place entries in the data
dictionary lock time . (very short
locking)

- BUILD : index rebuild changes journal table
. (no locking)

- MERGE : rebuild index journal table merge
. , journal table
(commit or rollback
) table lock journal



table and index online .
(waiting for transaction and locking for merge)

- Index online DML transaction
journal table index online option
parallel DML ,
bitmap/cluster/IOT secondary index .

- Index block block
Oracle8i coalesce command coalesces leaf blocks
within the same branch of the tree free block free
lists .

- Example

SQL> alter index xakemp_name_salcnt_desc rebuild online ;
Index altered.

SQL> create index xakemp_mgr on emp (mgr) online ;
Index created.

SQL> alter index XAKEMP_NAME_SALCNT_DESC coalesce ;
Index altered.

Computing statistics on indexes

- statistics generation create index and alter index
rebuild analyze .
create index …… compute statistics ;
alter index …….. rebuild compute statistics ;

- non-partitioned index : gather table, column, and index
statistics
concatenated index : refer only to the first column
(index composite leading column )
partitioned index : only index statistics .

- Example

SQL> alter index xakemp_name_salcnt_desc rebuild compute statistics ;
Index altered.

SQL> create index xakemp_mgrdept on emp (mgr, deptno)
2 compute statistics ;
Index created.

CF. test Index build or
rebuild online option compute statistics option
.

CF. rebuild reverse option online option
8108 error return . (… noreverse online ; )



Index-Organized Tables

Basic concept Oracle8 New Features

IOT & Logical ROWIDs(Oracle8i New Features)

- IOT index table , PK index value +
non-key value permanent physical
rowid . (leaf block row header, key column,
nonkey column ) Oracle8
secondary index Oracle8i
Logical Rowid concept .
CF. PK index rowid .
, PK value index .

SQL> create table emp_iot (
2 empno number not null,
3 ename varchar2(10),
4 deptno number,
7. table default 5 emp_desc varchar2(20),
organization heap 6 constraint xpkemp_iot primary key (empno))
8. 1 block 1 row 30% 7 organization index tablespace tools
nonkey 8 pctthreshold 30
9 overflow tablespace users ;
column data overflow
Table created
9. pctthreadshold
data tablespace SQL> select segment_name, segment_type, tablespace_name
( 2 from user_segments
user default tablespace 3 where segment_name like ('%IOT%') ;
.
SEGMENT_NAME SEGMENT_TYPE TABLESPACE_NAME
---------------------------------------------------------------------------------------------
SYS_IOT_OVER_40973 TABLE USERS
XPKEMP_IOT INDEX TOOLS

- user_segments view
table name emp_iot segment .
, pctthreshold overflow
SYS_IOT table type overflow segment
.

- Secondary Index :
PK column column index
data ? IOT
permanent physical rowid index
physical rowid rowid
. secondary index PK values
8i NF ‘Universal Rowid’ , Urowid
index .



- Data search using secondary index :
Physical Guess : indexing index statistics
physical rowid physical
rowid access .
Without guess or after incorrect guess : PK values
logical rowid access .
CF. UROWID compatible 8.1 .
CF. PL/SQL fully supports the ‘UROWID’ datatype.
SQL> explain plan for select ename from emp_iot
2 where deptno = 10 ;
Explained.

3 from plan_table
4 start with id=0
6 order by id;
Plan View
-------------------------------------------------------------------------------
SELECT STATEMENT
INDEX FAST FULL SCAN XPKEMP_IOT

SQL> create index xakempiot_deptno on emp_iot (deptno) ;
Secondary index Index created.
execution plan SQL> explain plan for select ename from emp_iot
. 2 where deptno = 10 ;
Explained.

3 from plan_table
4 start with id=0
6 order by id;
Plan View
-------------------------------------------------------------------------------
SELECT STATEMENT
INDEX UNIQUE SCAN XPKEMP_IOT
INDEX RANGE SCAN XAKEMPIOT_DEPTNO

- Unchanged Restrictions :
1. must have a primary key
2. cannot use unique constraints Snapshot
3. cannot be clustered M-View
4. cannot contain LONG columns .
5. cannot support distribution and replication
6. cannot support object types
7. can contain LOB columns and nested table types(varray), but
only if the table is not partitioned
8. IOT rowid datatype (rowid )
9. (like index rebuild command) can use move command for
reorganizing data(not overflow segment)
CF. IOT PK index rebuild secondary
index rebuild .



5. Basic Partition

HASH Partition

Oracle8 New Features range range
partition Oracle8i range partition
hash partition . concept Oracle8
.

tablespace (ts_emp1 ~ 6)
create tablespace command .

Partitioned key update

- update command partition key
( partition range value ) row movement
.

- : alter table table_name ENABLE row movement ;
default value DISABLE partitioned key update
ENABLE setting .

- Example

SQL> create table emp_part (
2 EMPNO NUMBER(4) not null,
3 ENAME VARCHAR2(10),
4 JOB VARCHAR2(9),
5 MGR NUMBER(4),
6 HIREDATE DATE,
7 SAL NUMBER(7,2),
8 COMM NUMBER(7,2),
9 DEPTNO NUMBER(2))
10 partition by range (hiredate)
11 ( partition p_emp_1980 values less than
12 (to_date('01-JAN-1981','DD-MON-YYYY')) tablespace ts_emp1,
ts_emp1 ~6 6 13 partition p_emp_1981 values less than
tablespace 14 (to_date('01-JAN-1982','DD-MON-YYYY')) tablespace ts_emp2,
hiredate range 6 15 partition p_emp_1982 values less than
partition partition
17 partition p_emp_1983 values less than
table emp 18 (to_date('01-JAN-1984','DD-MON-YYYY')) tablespace ts_emp4,
table data insert 19 partition p_emp_1984 values less than
. 20 (to_date('01-JAN-1985','DD-MON-YYYY')) tablespace ts_emp5,
22 (MAXVALUE) tablespace ts_emp6 ) ;
Table created.

SQL> insert into emp_part select * from emp ;
162034 rows created.

SQL> commit ;
Commit complete.



SQL> select distinct(to_char(hiredate, 'YYYY-MM-DD'))
2 from emp_part partition(p_emp_1980) ;
(TO_CHAR(H
Partition select1 ----------
data partition 1980-12-17
insert
. SQL> select distinct(to_char(hiredate, 'YYYY-MM-DD'))
2 from emp_part partition(p_emp_1985) ;
(TO_CHAR(H
----------
1985-10-01

SQL> update emp_part partition(p_emp_1985)
2 set empno = 0, hiredate = to_date('19801231', 'YYYYMMDD')
3 where rownum = 1 ;
update emp_part partition(p_emp_1985)
*
ERROR at line 1:
ORA-14402: updating partition key column would cause a partition change

Partition key update SQL> alter table emp_part enable row movement ;
enable row Table altered.
movement partition
key update SQL> update emp_part partition(p_emp_1985)
select 2 set empno = 0, hiredate = to_date('19801231', 'YYYYMMDD')
3 where rownum = 1 ;
.
1 row updated.

SQL> commit ;
Commit complete.

SQL> alter session set nls_date_format = 'YYYYMMDD' ;
Session altered.

SQL> select empno, ename, hiredate from emp_part partition(p_emp_1980)
2 where empno = 0 ;
EMPNO ENAME HIREDATE
---------- ---------- --------
0 LEE 19801231

Hash function

- table data hash function
partition . ( , I/O
)

- It controls the physical placement of data across a fixed
number of partitions and gives you a highly tunable method of
data placement.

- Does not support LONG, LONG RAW, ROWID, UROWID,
LOB datatypes

- Hash partition key column list 16 .



Hash partition

- partition data .
, Number of partitions should be a power of 2 (2, 4, 6, …)
(2 )
CF. test 2 error
. ,2
data
.
CF. 2 partition tablespace
tablespace partition
.

- parallel DML performance .

- deptno column value 6
tablespace 12 partition data
. tablespace
2 partition .

SQL> create table emp_hash_part (
4 JOB VARCHAR2(9),
5 MGR NUMBER(4),
6 HIREDATE DATE,
7 SAL NUMBER(7,2),COMM NUMBER(7,2),
8 DEPTNO NUMBER(2))
9 storage ( initial 64k next 64k pctincrease 0)
10 partition by hash (deptno) partitions 12
11 store in (ts_emp1, ts_emp2, ts_emp3, ts_emp4, ts_emp5, ts_emp6) ;
storage
Table created.
partition
hash partition disk SQL> set pagesize 100
stripping tablespace SQL> col table_name for a15
, . SQL> col partition_name for a15
SQL> col tablespace_name for a10
SQL> select table_name, PARTITION_NAME, TABLESPACE_NAME
2 from user_tab_partitions
Hash partition name 3 where table_name like '%HASH%'
default 4 order by tablespace_name ;
SYS_Pnn . TABLE_NAME PARTITION_NAME TABLESPACE
--------------- --------------- ----------
EMP_HASH_PART SYS_P19 TS_EMP1
12 rows selected.



- Range partition user partition
hash partition hash method Oracle
data .(no logical meaning to the users)
select command .
select * from PARTITION(part_name) where…….. ;
CF. test partition select
. , hash function
partition partition return
.
SQL> insert into emp_hash_part select * from emp ;

SQL> commit ;
Commit complete.

SQL> select count(*) from emp_hash_part partition (SYS_P19) ;
COUNT(*)
----------
46291

- tablespace partition hash
partition naming . Storage range
partition table level
storage partition and tablespace .

SQL> create table emp_name_hash_part (
4 HIREDATE DATE,
5 DEPTNO NUMBER(2))
6 storage (initial 32k)
7 partition by hash (hiredate)
8 ( partition hp_h1 tablespace ts_emp1,
9 partition hp_h2 tablespace ts_emp2,
10 partition hp_h3 tablespace ts_emp2)
Table created.
3 where table_name = 'EMP_NAME_HASH_PART'
4 order by tablespace_name ;

TABLE_NAME PARTITION_NAME TABLESPACE
-------------------- --------------- ----------
EMP_NAME_HASH_PART HP_H1 TS_EMP1

Partitioned index restriction
- Bitmap index on the partitioned table must be local.

- Global Index range partitioned table only. , hash
partitioned table global index .



6. Partition Maintenance

8i New Maintenance Function

Basic Partition Maintenance Oracle8 New Features
.

Add hash partition

- hash partition
CF. partitioned table
.

- partition name and tablespace
partition name SYS_Pnn tablespace
table’s default tablespace .

CF. Other attributes are always inherited from table-level defaults.
(add partition
table )

- Example
-------------------- --------------- ----------

SQL> insert into emp_name_hash_part
2 select empno, ename, hiredate, deptno from emp ;
Hash function
partition data SQL> commit ;
. SQL Commit complete.
SQL> select count(*) from emp_name_hash_part partition(hp_h1) ;
partition alias
COUNT(*)
----------
. 7718
COUNT(*)
----------
115728
SQL> select count(a.empno) from emp_name_hash_part partition(hp_h3) a ;
COUNT(A.EMPNO)
--------------
38588



SQL> alter table emp_name_hash_part add partition ;
Table altered.

Partition name tablespace
hash partition 4 order by tablespace_name ;
tablespace user default
tablespace tools TABLE_NAME PARTITION_NAME TABLESPACE
partition name SYS_P42 -------------------- --------------- ----------
. EMP_NAME_HASH_PART SYS_P42 TOOLS

COUNT(*)
----------
7718

COUNT(*)
----------
84865
Partition hp_h2 data
new partition sys_p42 COUNT(*)
----------
. 38588
Data
partition SQL> select count(*) from emp_name_hash_part partition(SYS_P42) ;
. COUNT(*)
----------
30863
SQL> alter table emp_name_hash_part add
2 partition hp_h4 tablespace ts_emp4 ;
Table altered.

Partition name tablespace 2 from user_tab_partitions
new partition 4 order by tablespace_name ;
.
-------------------- --------------- ----------
EMP_NAME_HASH_PART SYS_P42 TOOLS



COUNT(*)
----------
0

COUNT(*)
----------
84865
Partition
data hp_h1 SQL> select count(*) from emp_name_hash_part partition(hp_h3) ;
COUNT(*)
hp_h4
----------
. Hash function 38588

SQL> select count(*) from emp_name_hash_part partition(SYS_P42) ;
. COUNT(*)
----------
30863

COUNT(*)
----------
7718

- parallel partition data
redistribute .
data overhead
partition partition
redistribution .

- , data skew current partition
2 hash partition .

- Index add partition .

CF. , 2 hash partition 1 data
distribution rate 1:1 2:1:1 . ,
2 1:1:1:1 .

CF. test data hash function

.



Merge partition

- Oracle8 partition
merge . , .

- Range partition partition
merge . (hash partition or IOT )

- new tablespace and storage table
default .

- partition partition tablespace
partition new data
segment .

CF. new partition segment partition
drop tablespace
data .

- The new index is placed in the default tablespace specified at
the index level. (merge partition index unusable )
SQL> create index xakemp_part_hdt on emp_part (hiredate) local ;
Index created.

SQL> select index_name, partition_name, tablespace_name
2 from user_ind_partitions
3 where index_name = 'XAKEMP_PART_HDT'

INDEX_NAME PARTITION_NAME TABLESPACE
------------------------------ --------------- ----------
XAKEMP_PART_HDT P_EMP_1980 TS_EMP1
Test tablespace
local index
table index partition
table SQL> select table_name, PARTITION_NAME, TABLESPACE_NAME
partition tablespace 2 from user_tab_partitions
3 where table_name = 'EMP_PART'
. 4 order by tablespace_name ;

-------------------- --------------- ----------
EMP_PART P_EMP_1980 TS_EMP1



SQL> alter table emp_part merge partitions p_emp_1980, p_emp_1982
2 into partition p_emp_1980_2 ;
(adjacent) alter table emp_part merge partitions p_emp_1980, p_emp_1982
partition merge *
ERROR at line 1:
ORA-14274: partitions being merged are not adjacent

2 into partition p_emp_1980_1 ;
Table altered.

partition merge SQL> select table_name, PARTITION_NAME, TABLESPACE_NAME
user default tablespace TABLE_NAME PARTITION_NAME TABLESPACE
-------------------- --------------- ----------
. EMP_PART P_EMP_1980_1 TOOLS

2 into partition p_emp_1983_4 tablespace ts_emp2
3 storage ( initial 64k next 64k pctincrease 0 ) ;
Table altered.

Partition merge SQL> select table_name, PARTITION_NAME, TABLESPACE_NAME
new 2 from user_tab_partitions
merge partition 3 where table_name = 'EMP_PART'
. 4 order by tablespace_name ;

-------------------- --------------- ----------
EMP_PART P_EMP_1980_1 TOOLS
EMP_PART P_EMP_1983_4 TS_EMP2

3 where index_name = 'XAKEMP_PART_HDT'
Local index table 4 order by tablespace_name ;
merge
SQL> select index_name, partition_name, tablespace_name, status
.
Index 3 where index_name = 'XAKEMP_PART_HDT'
partition unusable 4 order by tablespace_name ;
. rebuild
. INDEX_NAME PARTITION_NAME TABLESPACE STATUS
------------------------------ --------------- ---------- --------
XAKEMP_PART_HDT P_EMP_1980_1 TOOLS UNUSABLE
XAKEMP_PART_HDT P_EMP_1983_4 TS_EMP2 UNUSABLE
XAKEMP_PART_HDT P_EMP_1982 TS_EMP3 USABLE



SQL> alter index xakemp_part_hdt rebuild partition
2 p_emp_1980_1 tablespace ts_emp1 ;
Index altered.

Rebuild partition command 3 where index_name = 'XAKEMP_PART_HDT'
index 4 order by tablespace_name ;

INDEX_NAME PARTITION_NAME TABLESPACE STATUS
------------------------ --------------- ---------- --------
XAKEMP_PART_HDT P_EMP_1980_1 TS_EMP1 USABLE
XAKEMP_PART_HDT P_EMP_1983_4 TS_EMP2 UNUSABLE

SQL> alter table emp_part modify partition p_emp_1983_4
2 rebuild unusable local indexes ;
Table altered.

Modify partition command 3 where index_name = 'XAKEMP_PART_HDT'
index 4 order by tablespace_name ;

------------------------ --------------- ---------- --------

Hash partition coalesce

- hash partition range partition merge
operation . (Removes a single hash
partition and redistributes the data)

- Hash function dissolution partition
drop and redistribution .(parallel )

- Merge local index rebuild .
SQL> create index xakempname_hash_hdt
2 on emp_name_hash_part (hiredate) local ;
Index created.
Test local index 2 from user_ind_partitions
index status 3 where index_name = 'XAKEMPNAME_HASH_HDT'
------------------------------ --------------- ---------- --------
XAKEMPNAME_HASH_HDT SYS_P42 TOOLS USABLE
XAKEMPNAME_HASH_HDT HP_H1 TS_EMP1 USABLE



SQL> alter table emp_name_hash_part add
2 partition hp_h5 tablespace ts_emp5 ;
Table altered.
-------------------- --------------- ----------
Hash partition hp_h5 EMP_NAME_HASH_PART SYS_P42 TOOLS
partition EMP_NAME_HASH_PART HP_H1 TS_EMP1
coalesce command EMP_NAME_HASH_PART HP_H2 TS_EMP2
hp_h5 partition EMP_NAME_HASH_PART HP_H3 TS_EMP2
. , hash function EMP_NAME_HASH_PART HP_H4 TS_EMP4
dissolution EMP_NAME_HASH_PART HP_H5 TS_EMP5
partition hp_h5
SQL> alter table emp_name_hash_part coalesce partition ;
hp_h2 index partition Table altered.
unusable data SQL> select table_name, PARTITION_NAME, TABLESPACE_NAME
redistribution hp_h2 2 from user_tab_partitions
. 3 where table_name = 'EMP_NAME_HASH_PART'
-------------------- --------------- ----------
EMP_NAME_HASH_PART SYS_P42 TOOLS

3 where index_name = 'XAKEMPNAME_HASH_HDT'
------------------------------ --------------- ---------- --------
XAKEMPNAME_HASH_HDT HP_H2 TS_EMP2 UNUSABLE

SQL> alter index xakempname_hash_hdt rebuild partition hp_h2 ;
Index altered.
Rebuild hp_h2 index 3 where index_name = 'XAKEMPNAME_HASH_HDT'
partition 4 order by tablespace_name ;

------------------------------ --------------- ---------- --------



General operation

- Oracle8 .

SQL> alter table emp_part rename partition p_emp_1983_4 to p_emp_old ;
Table altered.
Partition rename : table, index
SQL> alter index xakempname_hash_hdt rename
.
2 partition SYS_P42 to hp_h5 ;
Index altered.

Move partition : partition SQL> alter table emp_part move partition p_emp_1980_1
tablespace 2 tablespace ts_emp1 ;
Table altered.
.
SQL> alter table emp_part add partition p_emp_1986
2 values less than
Add partition : hash partition 3 (to_date('01-JAN-1987','DD-MON-YYYY')) tablespace ts_emp1 ;
range partition alter table emp_part add partition p_emp_1986
range value *
ERROR at line 1:
ORA-14074: partition bound must collate higher than that of the last partition
partition
. SQL> alter table emp_part split partition p_emp_1985
highest 2 at (to_date('01-JAN-1986','DD-MON-YYYY'))
partition range 3 into ( partition p_emp_1985 tablespace ts_emp4,
maxvalue error 4 partition p_emp_1986 tablespace ts_emp5) ;
return
split command Table altered.
.
SQL> set long 2000
SQL> select table_name, partition_name, high_value
4 order by 2 ;

TABLE_NAME PARTITION_NAME HIGH_VALUE
Split ----------------------------------------------------------------
p_emp_1986 range EMP_PART P_EMP_1980_1 TO_DATE('1982-01-01 00:00:00'...)
EMP_PART P_EMP_1982 TO_DATE('1983-01-01 00:00:00'...)
maxvalue
EMP_PART P_EMP_OLD TO_DATE('1985-01-01 00:00:00'...)
. EMP_PART P_EMP_1985 TO_DATE('1986-01-01 00:00:00'...)
EMP_PART P_EMP_1986 MAXVALUE

SQL> select index_name, partition_name, status from user_ind_partitions
2 where INDEX_NAME = 'XAKEMP_PART_HDT' ;
P_EMP_OLD rename
local index p_emp_1983_4 INDEX_NAME PARTITION_NAME STATUS
status . --------------- --------------- --------
XAKEMP_PART_HDT P_EMP_1983_4 USABLE
XAKEMP_PART_HDT P_EMP_1982 USABLE
XAKEMP_PART_HDT P_EMP_1980_1 UNUSABLE
Move or Split command XAKEMP_PART_HDT P_EMP_1985 UNUSABLE
local index XAKEMP_PART_HDT P_EMP_1986 USABLE
unusable .



SQL> alter table emp_part drop partition p_emp_old ;
Table altered.
Drop partition : table partition
drop local index SQL> select index_name, partition_name, status from user_ind_partitions
p_emp_1983_4 drop 2 where INDEX_NAME = 'XAKEMP_PART_HDT' ;
. Hash partitioned table
INDEX_NAME PARTITION_NAME STATUS
.
--------------- --------------- --------
XAKEMP_PART_HDT P_EMP_1980_1 UNUSABLE
XAKEMP_PART_HDT P_EMP_1985 UNUSABLE
Index partition drop XAKEMP_PART_HDT P_EMP_1986 USABLE
error return . ,
table partition drop local SQL> alter index xakemp_part_hdt drop partition p_emp_1980_1 ;
index partition drop alter index xakemp_part_hdt drop partition p_emp_1980_1
. global index *
drop index partition ERROR at line 1:
ORA-14076: DROP/SPLIT PARTITION may not be applied to a LOCAL
highest
index partition
partition drop .
SQL> alter table emp_part modify partition p_emp_1986
2 storage ( next 1m ) ;
Table altered.
Modify partition : table or
index partition storage SQL> alter index xakemp_part_hdt modify partition p_emp_1980_1
2 storage ( next 1m ) ;
.
Index altered.
, index global or
local partitioned index SQL> alter index xakempname_hash_hdt modify partition hp_h5
. 2 storage ( next 1m ) ;
alter index xakempname_hash_hdt modify partition hp_h5
Hash partition : *
hash partition ERROR at line 1:
storage table level ORA-14192: cannot modify physical index attributes of a Hash index
modify command partition
error return .

Modify default attributes

- Table or index default attribute
partition attribute .

- , partition
partition explicitly storage attribute
default attribute override .

- manual
hash partitioned table index default
attribute tablespace
test storage clause Oracle8.1.6.2
.



-
table
table
.
SQL> drop table emp_part ;
Table dropped.

4 JOB VARCHAR2(9),
5 MGR NUMBER(4),
6 HIREDATE DATE,
7 SAL NUMBER(7,2),
8 COMM NUMBER(7,2),
9 DEPTNO NUMBER(2))
Table created.

Index created.

SQL> drop table emp_name_hash_part ;
Table dropped.

SQL> create table emp_name_hash_part (
4 HIREDATE DATE,
5 DEPTNO NUMBER(2))
6 storage (initial 32k)
7 partition by hash (hiredate)
8 ( partition hp_h1 tablespace ts_emp1,
9 partition hp_h2 tablespace ts_emp2,
10 partition hp_h3 tablespace ts_emp3)
Table created.

SQL> create index xakempname_hash_hdt
2 on emp_name_hash_part (hiredate) local ;
Index created.

SQL> break on table_name



- table attribute modify
.
SQL> select table_name, partition_name, max_extent
3 where table_name in ('EMP_PART', 'EMP_NAME_HASH_PART')
4 order by 1 ;

TABLE_NAME PARTITION_NAME MAX_EXTENT
-------------------- --------------- ------------------------------------------------
EMP_NAME_HASH_PART HP_H2 505
HP_H3 505
HP_H1 505
EMP_PART P_EMP_1981 505
P_EMP_1982 505
P_EMP_1980 505
P_EMP_1985 505
P_EMP_1983 505
P_EMP_1984 505

SQL> alter table emp_part modify default attributes
2 storage ( maxextents 1024 ) ;
Table default attribute Table altered.
maxextents 1024
. SQL> alter table emp_name_hash_part modify default attributes
2 storage ( maxextents 1024 ) ;
Table altered.

4 order by 1 ;

-------------------- --------------- ----------
HP_H3 505
partition HP_H1 505
max extent EMP_PART P_EMP_1981 505
.( 505 ) P_EMP_1982 505
P_EMP_1980 505
P_EMP_1985 505
P_EMP_1983 505
P_EMP_1984 505

SQL> alter table emp_part split partition p_emp_1985
2 at (to_date('01-JAN-1987','DD-MON-YYYY'))
3 into ( partition p_emp_1985 tablespace ts_emp1,
4 partition p_emp_1987 tablespace ts_emp2) ;
Table altered.

- emp_part p_emp_1985 partition
.



Split command 3 where table_name = 'EMP_PART';
partition
default attribute -------------------- --------------- ----------
. , split EMP_PART P_EMP_1981 505
partition new P_EMP_1982 505
partition P_EMP_1983 505
. P_EMP_1984 505
P_EMP_1985 505
P_EMP_1980 505
P_EMP_1986 505

SQL> alter table emp_part drop partition p_emp_1986 ;
Table altered.
Maxvalue partition
drop new SQL> alter table emp_part add partition p_emp_1986
partition hash 2 values less than ( maxvalue ) tablespace ts_emp6 ;
partitioned table add Table altered.
partition new
partition . SQL> alter table emp_name_hash_part add partition ;
Table altered.

4 order by 1 ;

-------------------- --------------- ----------
HP_H3 505
SYS_P47 1024
partition HP_H1 505
default attribute EMP_PART P_EMP_1981 505
P_EMP_1982 505
.
P_EMP_1986 1024
P_EMP_1980 505
P_EMP_1985 505
P_EMP_1983 505
P_EMP_1984 505



- index default attribute test extent
.

- hash partitioned index
.
SQL> break on index_name
SQL> select index_name, partition_name, next_extent
3 where index_name in
4 ('XAKEMP_PART_HDT','XAKEMPNAME_HASH_HDT')
5 order by 1 ;

INDEX_NAME PARTITION_NAME NEXT_EXTENT
------------------------------ --------------- -----------
XAKEMPNAME_HASH_HDT HP_H1 40960
HP_H2 40960
HP_H3 40960
SYS_P47 40960
XAKEMP_PART_HDT P_EMP_1981 40960
P_EMP_1982 40960
P_EMP_1980 40960
P_EMP_1985 40960
P_EMP_1983 40960
P_EMP_1984 40960
P_EMP_1986 40960

SQL> alter index xakemp_part_hdt modify default attributes
2 maxtrans 10 storage ( next 5M ) ;
index partition Index altered.
default attribute
. SQL> alter index xakempname_hash_hdt modify default attributes
2 maxtrans 10 storage ( next 5M ) ;
Index altered.

SQL> select index_name, partition_name, next_extent
3 where index_name in
5 order by 1 ;

INDEX_NAME PARTITION_NAME NEXT_EXTENT
------------------------------ --------------- -----------
XAKEMPNAME_HASH_HDT HP_H1 40960
HP_H2 40960
partition HP_H3 40960
attribute SYS_P47 40960
. XAKEMP_PART_HDT P_EMP_1981 40960
P_EMP_1982 40960
P_EMP_1980 40960
P_EMP_1985 40960
P_EMP_1983 40960
P_EMP_1984 40960
P_EMP_1986 40960



SQL> alter table emp_part drop partition p_emp_1986 ;
Table altered.

SQL> alter table emp_part add partition p_emp_1986
2 values less than ( maxvalue ) tablespace ts_emp6 ;
Table altered.

SQL> alter table emp_name_hash_part add partition ;
Table altered.
Add partition
local index next extent SQL> select index_name, partition_name, next_extent
default attribute 5M 2 from user_ind_partitions
. 3 where index_name in
5 order by 1 ;
partition default
attribute INDEX_NAME PARTITION_NAME NEXT_EXTENT
. ------------------------------ --------------- -----------
XAKEMPNAME_HASH_HDT SYS_P48 5242880
HP_H1 40960
HP_H3 40960
SYS_P47 40960
HP_H2 40960
XAKEMP_PART_HDT P_EMP_1981 40960
P_EMP_1986 5242880
P_EMP_1980 40960
P_EMP_1985 40960
P_EMP_1982 40960
P_EMP_1983 40960
P_EMP_1984 40960



Unusable local index data insert

- local index status unusable partition
data .
data insert index
.

- alter session command
.

- , index unique index
index partition query
error .

SQL> col index_name for a15
SQL> col partition_name for a15
SQL> select index_name, partition_name, status
3 where index_name = 'XAKEMP_PART_HDT' ;

--------------- --------------- --------


SQL> commit ;
Commit complete.

SQL> alter table emp_part move partition
2 p_emp_1981 tablespace tools ;
Table altered.
Data insert data
partition SQL> select index_name, partition_name, status
p_emp_1981 2 from user_ind_partitions
tablespace move 3 where index_name = 'XAKEMP_PART_HDT' ;
local index status unusable
. INDEX_NAME PARTITION_NAME STATUS
--------------- --------------- --------



SQL> insert into emp_part select * from emp
2 where to_char(hiredate, 'YYYYMMDD') like '1981%'
Move partition 3 and rownum = 1 ;
p_emp_1981 1980 insert into emp_part select * from emp
data insert error *
ERROR at line 1:
.
ORA-01502: index 'SCOTT.XAKEMP_PART_HDT' or partition of such
index unusable index is in
. unusable state

SQL> alter session set skip_unusable_indexes = true ;
Session altered.
Alter session command SQL> insert into emp_part select * from emp
data 2 where to_char(hiredate, 'YYYYMMDD') like '1981%'
insert 3 and rownum = 1 ;
. 1 row created.

SQL> commit ;
Commit complete.

SQL> alter table emp_part modify partition
2 p_emp_1981 rebuild unusable local indexes ;
Table altered.

index rebuild 3 where index_name = 'XAKEMP_PART_HDT' ;
index .
--------------- --------------- --------



Review command

- general operation
Oracle8 partition
Oracle8 sample
exchange command truncate
command .

- table creation
.
SQL> drop table emp_part ;
Table dropped.

4 JOB VARCHAR2(9),
5 MGR NUMBER(4),
6 HIREDATE DATE,
7 SAL NUMBER(7,2),
8 COMM NUMBER(7,2),
9 DEPTNO NUMBER(2))
Table created.

Index created.

Table test SQL> select count(*) from emp ;
1981 COUNT(*)
data ----------
162034
77161 .
data SQL> select count(*) from emp
partitioned table insert 2 where to_char(hiredate, 'YYYYMMDD') like '1981%' ;
test . COUNT(*)
----------
77161

SQL> commit ;
Commit complete.



SQL> col table_name for a10
SQL> col index_name for a15
SQL> col partition_name for a10
SQL> col tablespace_name for a10
SQL> set pagesize 100
SQL> select table_name, partition_name, tablespace_name
3 where table_name = 'EMP_PART' ;

TABLE_NAME PARTITION_ TABLESPACE
---------- ---------- ----------
data
partition p_emp_1981 select EMP_PART P_EMP_1985 TS_EMP6
emp 1981 EMP_PART P_EMP_1980 TS_EMP1

. partition SQL> select count(*) from emp_part partition(p_emp_1981) ;
.
COUNT(*)
----------
77161

SQL> col segment_name for a10
SQL> select segment_name, partition_name, extents
2 from user_segments
3 where segment_name = 'EMP_PART' ;

SEGMENT_NA PARTITION_ EXTENTS
---------- ---------- ----------
Partition truncate 3 EMP_PART P_EMP_1985 7
. Truncate EMP_PART P_EMP_1980 5
partition data
. SQL> alter table emp_part truncate partition p_emp_1981
2 drop storage ;
Table truncated.

SQL> select count(*) from emp_part partition(p_emp_1981) ;

COUNT(*)
----------
0

SQL> alter table emp_part truncate partition p_emp_1982
2 reuse storage ;
Table truncated.

SQL> alter table emp_part truncate partition p_emp_1980 ;
Table truncated.



Partition truncate drop 2 from user_segments
storage 3 where segment_name = 'EMP_PART' ;
option
extent 1 space ---------- ---------- ----------
reuse option EMP_PART P_EMP_1981 1
p_emp_1982 EMP_PART P_EMP_1982 7
partition storage EMP_PART P_EMP_1983 5
. EMP_PART P_EMP_1984 8

SQL> create table emp_history as select * from emp
2 where to_char(hiredate, 'YYYYMMDD') like '1981%' ;
Table created.
Exchange partition test
emp table SQL> select count(*) from emp_history ;
emp_history table
. table COUNT(*)
data 1981 ----------
. 77161

SQL> alter table emp_part exchange partition p_emp_1981
2 with table emp_history ;
Table altered.

SQL> select count(*) from emp_history ;

truncate p_emp_1981 COUNT(*)
partition emp_history ----------
0
exchange table
partition SQL> select count(*) from emp_part partition(p_emp_1981) ;
.
, physical segment COUNT(*)
. ----------
77161


---------- ---------- ----------
EMP_PART P_EMP_1981 TOOLS



3 where segment_name = 'EMP_PART' ;

---------- ---------- ----------
Exchange data EMP_PART P_EMP_1983 5
p_emp_1981 partition EMP_PART P_EMP_1980 1
extents . SQL> select index_name, partition_name, status
index unusable 2 from user_ind_partitions
. 3 where index_name = 'XAKEMP_PART_HDT' ;

------------------- --------------- --------
SQL> alter table emp_part move partition p_emp_1981
2 tablespace ts_emp2 ;
Table altered.

---------- ---------- ----------
Partition move rebuild EMP_PART P_EMP_1983 TS_EMP4
partitioned table EMP_PART P_EMP_1984 TS_EMP5
table EMP_PART P_EMP_1985 TS_EMP6
. EMP_PART P_EMP_1980 TS_EMP1
SQL> alter table emp_part modify partition
2 p_emp_1981 rebuild unusable local indexes ;
Table altered.
3 where index_name = 'XAKEMP_PART_HDT' ;

------------------------------ --------------- --------
emp_history drop XAKEMP_PART_HDT P_EMP_1985 USABLE
. XAKEMP_PART_HDT P_EMP_1980 USABLE

SQL> drop table emp_history ;
Table dropped.



Execution PLAN : ITERATOR and INLIST

- emp_part data
.

- partitioned table emp_part hiredate
partition key . 6 partition
1980 1985 data key .

- hiredate column local prefixed index
.

- hiredate SQL Oracle8i
execution plan

SQL> truncate table emp_part ;
Table truncated.
SQL> commit ;
Commit complete.
SQL> alter session set nls_date_format = 'YYYYMMDD HH24:MI:SS' ;
Session altered.
SQL> select distinct(hiredate) from emp_part ;
HIREDATE
test emp table Oracle -----------------
emp table 19801217 00:00:00
19810220 00:00:00
row
19810222 00:00:00
insert sample 19810402 00:00:00
162034 row hiredate 19810501 00:00:00
15 19810609 00:00:00
. 19810908 00:00:00
19810928 00:00:00
19811117 00:00:00
19811203 00:00:00
19820123 00:00:00
19821209 00:00:00
19830112 00:00:00
19840918 00:00:00
19851001 00:00:00
SQL> set pagesize 100
SQL> set long 2000
SQL> select table_name, partition_name, high_value
4 order by 2 ;

TABLE_NAME PARTITION_NAME HIGH_VALUE
----------------------------------------------------------
EMP_PART P_EMP_1980 TO_DATE(' 1981-01-01 00:00:00'…
EMP_PART P_EMP_1985 MAXVALUE



SQL> col operation format a16
SQL> col options format a20
SQL> col start format a12
SQL> col stop format a12
SQL> col distribution format a10
SQL> col object_name format a15
SQL> set numwidth 7

SQL> select distinct ename, hiredate from emp_part
partition 2 where hiredate = to_date('19810501 00:00:00', 'YYYYMMDD HH24:MI:SS') ;
data select . ENAME HIREDATE
---------- --------
BLAKE 19810501

SQL> explain plan for select distinct ename, hiredate
2 from emp_part
3 where hiredate = to_date('19810501 00:00:00', 'YYYYMMDD HH24:MI:SS') ;
Explained.

SQL> select operation, options, object_name,
2 partition_start "start", partition_stop "stop",
Index 2 3 bytes, cardinality as "#"
partition data 4 from plan_table
return execution 5 order by id ;
plan .
OPERATION OPTIONS OBJECT_NAME start stop BYTES #
------------------------------------------------------------------------------------------------------------------
SELECT STATEMENT 16 1
SORT UNIQUE 16 1
TABLE ACCESS BY LOCAL INDEX ROWID EMP_PART 2 2 16 1
INDEX RANGE SCAN XAKEMP_PART_HDT 2 2 1

SQL> select distinct ename, hiredate from emp_part
2 where hiredate in
3 ( to_date('19810501 00:00:00', 'YYYYMMDD HH24:MI:SS'),
4 to_date('19840918 00:00:00', 'YYYYMMDD HH24:MI:SS')) ;

ENAME HIREDATE
---------- --------
partition BLAKE 19810501
partition data JANG 19840918
JUN 19840918
select .
KIM 19840918
LEE 19840918

SQL> delete from plan_table ;
10 rows deleted.

SQL> commit ;
Commit complete.



SQL> explain plan for select distinct ename, hiredate
2 from emp_part
3 where hiredate in
4 ( to_date('19810501 00:00:00', 'YYYYMMDD HH24:MI:SS'),
5 to_date('19840918 00:00:00', 'YYYYMMDD HH24:MI:SS')) ;
Explained.
Execution plan
KEY(INLIST) SQL> select operation, options, object_name,
select “in” 2 partition_start "start", partition_stop "stop",
partition 3 bytes, cardinality as "#"
. 4 from plan_table
5 order by id ;
, partition
partition access OPERATION OPTIONS OBJECT_NAME start stop BYTES #
. --------------------------------------------------------------------------------------------------------------------------------------------------
SELECTSTATEMENT 80 5
INLIST ITERATOR
PARTITION RANGE ITERATOR KEY(INLIST) KEY(INLIST)
SORT UNIQUE 80 5
TABLE ACCESS BY LOCAL INDEX ROWID EMP_PART KEY(INLIST) KEY(INLIST) 80 5
INDEX RANGE SCAN XAKEMP_PART_HDT KEY(INLIST) KEY(INLIST)

- Oracle8 version “in”
partition 2
partition partition
range access . Oracle8 Test
ITERATOR
KEY(INLIST) Oracle8i
.

- full table scan
( hiredate index )
Oracle8 “in” partition
partition ITERATOR( ) range
scan(partition full scan) (“in”
table full scan )
Oracle8i INLIST partition full scan
. , full
scan scan
.



Export

- export table partition export
query data export
.

- filesize option export dump file
dump file import
.

CF. FILESIZE option defatul(B:bytes), K(Kbytes), M(Mbytes),
G(Gbytes) .

- .
SQL> select count(*) from emp_part partition(p_emp_1981) ;

COUNT(*)
emp_part partition --------
p_emp_1981 data 77161 77161
deptno 20 SQL> select count(*) from emp_part partition(p_emp_1981)
15441 . 2 where deptno = 20 ;

COUNT(*)
--------
15441
$ exp scott/tiger file=emp1980_1.dmp tables=emp_part:p_emp_1981
query='"where deptno=20"' filesize=256K
Oracle8i new export option
partition, query, filesize Export: Release 8.1.6.2.0 - Production on Mon Aug 13 16:42:28 2001
export
. (c) Copyright 1999 Oracle Corporation. All rights reserved.

Connected to: Oracle8i Enterprise Edition Release 8.1.6.2.0 - Production
With the Partitioning option
JServer Release 8.1.6.2.0 - Production
Export done in US7ASCII character set and US7ASCII NCHAR character se
About to export specified tables via Conventional Path ...
. . exporting table EMP_PART
. . exporting partition P_EMP_1981
emp1980_1.dmp Export file: expdat.dmp > emp1980_2.dmp
256K
interactive mode continuing export into file emp1980_2.dmp
export dump filename
. 256K data Export file: expdat.dmp > emp1980_3.dmp
export
continuing export into file emp1980_3.dmp
.
15441 rows exported
ls command Export terminated successfully without warnings.
export file size 256K
. $ ls -l *.dmp
-rw-r--r-- 1 oracle dba 262144 Aug 13 16:42 emp1980_1.dmp



7. Composite Partitioning

Subpartition

Composite partition

- range partition + hash partition

- Composite partition range partition hash
function range partition hash partition
. hash partition
subpartition .

- Range high availability and manageability data
placement performance for
parallel DML and supports partition-wise joins
. (hash partition )

- physical data segment range partition
subpartition(hash partition)
. partition segment range
partition subpartition .

CF. range partition 4 subpartition 4 16
subpartition segment .

- partition segment .
SQL> alter session set nls_date_format = 'DD-MON-YYYY' ;
Session altered.

alter session command SQL> create table emp_comp_part (
range 2 EMPNO NUMBER(4) not null,
partition key date type 3 ENAME VARCHAR2(10),
partition 4 JOB VARCHAR2(9),
5 MGR NUMBER(4),
. Subpartition
6 HIREDATE DATE,
empno 4 7 SAL NUMBER(7,2),
range partition 8 COMM NUMBER(7,2),
tablespace 9 DEPTNO NUMBER(2))
composite partitioning 10 partition by range (hiredate)
. 11 subpartition by hash(empno) subpartitions 4
13 ('01-JAN-1981') tablespace ts_emp1,

Table created.



SQL> select segment_name, partition_name, tablespace_name
user_segments query 3 where segment_NAME = 'EMP_COMP_PART'
tablespace 4 order by 3 ;
subpartition
. SEGMENT_NAME PARTITION_NAME TABLESPACE
4 X 4 = 16 subpartitions --------------- --------------- ----------
EMP_COMP_PART SYS_SUBP94 TS_EMP1

SQL> create table emp_name_comp_part (
range partition 2 EMPNO NUMBER(4) not null,
subpartition
4 JOB VARCHAR2(9),
attribute 5 MGR NUMBER(4),
control 6 HIREDATE DATE,
. 7 SAL NUMBER(7,2),
8 COMM NUMBER(7,2),
9 DEPTNO NUMBER(2))
11 subpartition by hash(empno) subpartitions 4
12 store in (ts_emp1, ts_emp2, ts_emp3, ts_emp4)
14 ('01-JAN-1981') storage(maxextents 512),
16 ('01-JAN-1982') pctfree 40,
18 ('01-JAN-1983') store in (ts_emp5),
20 (MAXVALUE) subpartitions 6 store in (ts_emp6)) ;
Table created.

user_tab_partitions range SQL> select table_name, partition_name, tablespace_name
partition query ,4 2 from user_tab_partitions
range partition 3 where table_name = 'EMP_NAME_COMP_PART'
4 order by 3 ;
logical
structure . TABLE_NAME PARTITION_NAME TABLESPACE
-------------------- --------------- ----------
EMP_NAME_COMP_PART P_EMP_1980 TOOLS



user_segments query 3 where segment_NAME = 'EMP_NAME_COMP_PART'
composite partition 4 order by 3 ;
emp_name_comp_part table
18 hash partition SEGMENT_NAME PARTITION_NAME TABLESPACE
segment -------------------- --------------- ----------
. , composite EMP_NAME_COMP_PART SYS_SUBP109 TS_EMP1
EMP_NAME_COMP_PART SYS_SUBP113 TS_EMP1
partition hash partition
data . EMP_NAME_COMP_PART SYS_SUBP114 TS_EMP2
(4 X 2) + EMP_NAME_COMP_PART SYS_SUBP115 TS_EMP3
4 + EMP_NAME_COMP_PART SYS_SUBP112 TS_EMP4
6 = 18 subpartitions EMP_NAME_COMP_PART SYS_SUBP116 TS_EMP4

SQL> insert into emp_name_comp_part select * from emp ;

SQL> commit ;
emp table data Commit complete.
p_emp_1984 partition SQL> select count(*)
61721 row data 2 from emp_name_comp_part partition(p_emp_1984) ;
3 hiredate COUNT(*)
. p_emp_1984 ----------
subpartition sys_subp125 61721
hiredate
. SQL> select distinct hiredate
2 from emp_name_comp_part partition(p_emp_1984) ;
HIREDATE
-----------
12-JAN-1983
18-SEP-1984
01-OCT-1985

SQL> select distinct hiredate
2 from emp_name_comp_part subpartition(SYS_SUBP125) ;
HIREDATE
-----------
subpartition alias 12-JAN-1983
DML
SQL> update emp_name_comp_part subpartition(SYS_SUBP125) es
. 2 set es.hiredate = sysdate ;
subpartition 7713 rows updated.
.



SQL> select distinct hiredate
2 from emp_name_comp_part subpartition(SYS_SUBP125) ;

HIREDATE
-----------
14-AUG-2001

update SQL> commit ;
subpartition hiredate Commit complete.

range partition p_emp_1984 SQL> select distinct hiredate
hiredate 1983 data 2 from emp_name_comp_part partition(p_emp_1984) ;
2001 data
HIREDATE
. -----------
18-SEP-1984
01-OCT-1985
14-AUG-2001

SQL> create table emp_name_sub_part (
3 HIREDATE DATE,
4 SAL NUMBER(7,2),
5 DEPTNO NUMBER(2))
7 subpartition by hash(empno) subpartitions 4
8 store in (ts_emp1, ts_emp2, ts_emp3, ts_emp4)
9 ( partition p_emp_1980 values less than ('01-JAN-1981'),
10 partition p_emp_1981 values less than ('01-JAN-1982'),
11 partition p_emp_1982 values less than (MAXVALUE)
12 ( subpartition sp_1982_h1 tablespace ts_emp5,
13 subpartition sp_1982_h2 tablespace ts_emp6)) ;

Table created.
composite SQL> select segment_name, partition_name, tablespace_name
partition hash partition 2 from user_segments
3 where segment_NAME = 'EMP_NAME_SUB_PART'
. 4 order by 3 ;

SEGMENT_NAME PARTITION_NAME TABLESPACE
-------------------- --------------- ----------
EMP_NAME_SUB_PART SYS_SUBP127 TS_EMP1
EMP_NAME_SUB_PART SP_1982_H1 TS_EMP5
EMP_NAME_SUB_PART SP_1982_H2 TS_EMP6



Composite partitioned indexes

- composite partitioned indexes always local index
tablespace default table subpartition
tablespace .

- , composite partitioned table range partitioned
global index . composite partitioned
global index . (hash syntax .)
3 where table_name = 'EMP_NAME_SUB_PART' ;

composite partitioned TABLE_NAME PARTITION_NAME TABLESPACE
table range partition -------------------- --------------- ----------
logical structure EMP_NAME_SUB_PART P_EMP_1980 TOOLS
EMP_NAME_SUB_PART P_EMP_1981 TOOLS
. table local
EMP_NAME_SUB_PART P_EMP_1982 TOOLS
index index
. SQL> create index xakempnmsubpt_hdempno
2 on emp_name_sub_part (hiredate, empno) local ;
Index created.

local index partition SQL> select index_name, partition_name, tablespace_name, status
tablespace 2 from user_ind_partitions
logical structure 3 where index_name = 'XAKEMPNMSUBPT_HDEMPNO' ;
.
composite partitioned INDEX_NAME PARTITION_NAME TABLESPACE STATUS
--------------------- --------------- ---------- --------
local index STATUS
XAKEMPNMSUBPT_HDEMPNO P_EMP_1980 N/A
“N/A” .

3 where segment_NAME = 'XAKEMPNMSUBPT_HDEMPNO'
4 order by 3 ;

local index SEGMENT_NAME PARTITION_NAME TABLESPACE
--------------------- --------------- ----------
underlying table XAKEMPNMSUBPT_HDEMPNO SYS_SUBP127 TS_EMP1
subpartition XAKEMPNMSUBPT_HDEMPNO SYS_SUBP131 TS_EMP1
XAKEMPNMSUBPT_HDEMPNO SYS_SUBP128 TS_EMP2
. XAKEMPNMSUBPT_HDEMPNO SYS_SUBP129 TS_EMP3
XAKEMPNMSUBPT_HDEMPNO SP_1982_H1 TS_EMP5
XAKEMPNMSUBPT_HDEMPNO SP_1982_H2 TS_EMP6



CF. user_ind_partitions index status “USABLE”
“N/A” bug Oracle8.1.7
Oracle8i
dictionary( ) subpartition
“N/A” subpartition
subpartition “UNUSABLE”
user_ind_partitions
.

CF. , composite partition
subpartitions dictionary
.

SQL> select index_name, partition_name, subpartition_name, status
2 from user_ind_subpartitions
3 where index_name = 'XAKEMPNMSUBPT_HDEMPNO' ;

INDEX_NAME PARTITION_ SUBPARTITION_NA STATUS
----------------------- ---------- --------------- --------
XAKEMPNMSUBPT_HDEMPNO P_EMP_1980 SYS_SUBP127 USABLE
XAKEMPNMSUBPT_HDEMPNO P_EMP_1982 SP_1982_H1 USABLE
XAKEMPNMSUBPT_HDEMPNO P_EMP_1982 SP_1982_H2 USABLE

SQL> select table_name, partition_name, subpartition_name, tablespace_name
2 from user_tab_subpartitions
3 where table_name = 'EMP_NAME_SUB_PART' ;

TABLE_NAME PARTITION_ SUBPARTITION_NA TABLESPACE
-------------------- ---------- --------------- ----------
EMP_NAME_SUB_PART P_EMP_1980 SYS_SUBP127 TS_EMP1
EMP_NAME_SUB_PART P_EMP_1982 SP_1982_H1 TS_EMP5
EMP_NAME_SUB_PART P_EMP_1982 SP_1982_H2 TS_EMP6

.



Different partition maintenance operation

- subpartition operation subpartition
.

- Alter table modify partition add/coalesce subpartition …
Alter table/index modify subpartition …
Alter index rebuild subpartition …
Alter table modify subpartition…
Alter table/index rename subpartition…
Alter table move subpartition…
Alter table exchange subpartition…
Alter table truncate subpartition…
Alter table modify subpartition unusable local indexes…
Alter table modify subpartition rebuild unusable local
indexes.

- composite partitioned table
subpartition default attributes .
SQL> alter table emp_name_sub_part
2 modify default attributes for partition p_emp_1982
3 tablespace ts_emp6 ;
Table altered.
CF. partition .

New dictionary and columns

- DBA/USER/ALL_TAB_SUBPARTITIONS : subpartition
information

- DBA/USER/ALL_SUBPART_KEY_COLUMNS : subpartition
key information

- DBA/USER/ALL_IND_SUBPARTITIONS : subpartition index
information

- DBA/USER/ALL_SUBPART_COL_STATISTICS : subpartition
column statistics

- DBA/USER/ALL_SUBPART_HISTOGRAMS : subpartition
histogram

- object data dictionary subpartition
columns .

Others

- partition subpartition subpartition
export/import . (6 export )

- subpartition analyze .
SQL> analyze table emp_name_sub_part subpartition
(sp_1982_h1) compute statistics ;



8. Added New Features for partition

Large Objects

Partition

- CF

Part/Type LOB Object Type IOT

RANGE O O O

HASH O O X

COMPOSITE O O X

CF. collection type IOT partition
IOT partitioning partition key primary key
column subset .

LOB (Large Objects)

- LOB column partition key LOB
column table partitioning . , LOB
data table partition keys partition .

- Lob partition tablespace
“tablespace default” table partition
tablespace . , table partition
tablespace table partition user
default tablespace lob partition
.

- Lob index lob partition
tablespace . lob index storage
.

- Alter command LOB storage
tablespace move command
. Move command lob partition
table partition .
table partition move lob partition
.

- Lob data partitioned table tablespace
“drop tablespace including contents” drop
all table partition tablespace ,
tablespace LOB partition



partition drop . ,1 table
partition drop tablespace
tablespace ORA-22868 error return
.

- LOB partition
.
SQL> create table emp_lob (
4 HIREDATE DATE,
5 SAL BLOB,
6 DEPTNO NUMBER(2))
9 (to_date('01-JAN-1985','DD-MON-YYYY')) tablespace ts_emp1
10 LOB(sal) store as (tablespace ts_emp6),
12 (MAXVALUE) tablespace ts_emp2
13 LOB(sal) store as (tablespace ts_emp5)) ;

Table created.

3 where table_name = 'EMP_LOB' ;
LOB column table
lob partition TABLE_NAME PARTITION_NA TABLESPACE
lob ---------- ------------ ----------
tablespace lob EMP_LOB P_EMP_1984 TS_EMP1
partition EMP_LOB P_EMP_1985 TS_EMP2
.
SQL> select table_name, partition_name, lob_partition_name, tablespace_name
Lob partition 2 from user_lob_partitions
system generated name 3 where table_name = 'EMP_LOB' ;
SYS_LOB_Pnn
. TABLE_NAME PARTITION_NA LOB_PARTITIO TABLESPACE
---------- ------------ ------------ ----------
EMP_LOB P_EMP_1984 SYS_LOB_P137 TS_EMP6
EMP_LOB P_EMP_1985 SYS_LOB_P138 TS_EMP5

SQL> create table emp_lob2 (
4 HIREDATE DATE,
5 SAL BLOB,
Table level lob 6 DEPTNO NUMBER(2))
7 LOB(sal) store as (tablespace default)
partition level
lob 9 ( partition p_emp_1984 values less than
partitioned table 10 (to_date('01-JAN-1985','DD-MON-YYYY')) tablespace ts_emp3
. 11 LOB(sal) store as (tablespace ts_emp4),
12 partition p_emp_1985 values less than (MAXVALUE)) ;

Table created.



3 where table_name = 'EMP_LOB2' ;

TABLE_NAME PARTITION_NA TABLESPACE
---------- ------------ ----------
EMP_LOB2 P_EMP_1984 TS_EMP3
EMP_LOB2 P_EMP_1985 TOOLS

Partition level SQL> select table_name, partition_name, lob_partition_name, tablespace_name
p_emp_1985 2 from user_lob_partitions
user default tablespace 3 where table_name = 'EMP_LOB2' ;
tools
table level lob TABLE_NAME PARTITION_NA LOB_PARTITIO TABLESPACE
---------- ------------ ------------ ----------
tablespace default
EMP_LOB2 P_EMP_1984 SYS_LOB_P141 TS_EMP4
lob partition EMP_LOB2 P_EMP_1985 SYS_LOB_P142 TOOLS
table partition
. SQL> alter table emp_lob2 move partition
2 p_emp_1985 tablespace ts_emp5 ;

Table altered.


---------- ------------ ----------

2 from user_lob_partitions
Move partition 3 where table_name = 'EMP_LOB2' ;
partition lob
partition . TABLE_NAME PARTITION_NA LOB_PARTITIO TABLESPACE
lob partition ---------- ------------ ------------ ----------
move EMP_LOB2 P_EMP_1984 SYS_LOB_P141 TS_EMP4
. EMP_LOB2 P_EMP_1985 SYS_LOB_P142 TOOLS

SQL> alter table emp_lob2 move partition p_emp_1985
2 lob(sal) store as (tablespace ts_emp5) ;

Table altered.


TABLE_NAME PARTITION_NA LOB_PARTITIO TABLESPACE
---------- ------------ ------------ ----------



SQL> alter table emp_lob2 move partition p_emp_1984
2 tablespace ts_emp1 lob(sal) store as (tablespace ts_emp1) ;
Table altered.

Move partition table SQL> select table_name, partition_name, tablespace_name
partition lob partition
.
---------- ------------ ----------


TABLE_NAME PARTITION_NA LOB_PARTITIO TABLESPACE
---------- ------------ ------------ ----------

SQL> create index xakemplob2_hdtemp on emp_lob2
2 (hiredate, empno) local ;
Index lob index
Index created.
partition lob partition
SQL> select table_name, partition_name, lob_partition_name,
. Lob index partition 2 lob_indpart_name, tablespace_name
name SYS_IL_Pnn 3 from user_lob_partitions
. 4 where table_name = 'EMP_LOB2';

TABLE_NAME PARTITION_NA LOB_PARTITIO LOB_INDPART_NAM
TABLESPACE
---------- ------------ ------------ --------------- ----------
EMP_LOB2 P_EMP_1985 SYS_LOB_P145 SYS_IL_P146 TS_EMP5
Table partition EMP_LOB2 P_EMP_1984 SYS_LOB_P147 SYS_IL_P148 TS_EMP1
lob partition
tablespace SQL> connect internal
tablespace drop Connected.
tablespace SQL> drop tablespace ts_emp1 including contents ;
table partition drop tablespace ts_emp1 including contents
*
tablespace drop
ERROR at line 1:
. ORA-22868: table with LOBs contains segments in different tablespaces

LOB Operation

- lob partition move
partition partition lob
.

- , lob storage lob .
SQL> alter table emp_lob2 modify lob (sal) (storage (next 1m
maxextents 512)) ;



New/Changed dictionary and columns

- DBA/USER/ALL_PART_LOBS : table level lob
information

- DBA/USER/ALL_LOB_PARTITIONS : lob partition
information (including lob index partition)

- DBA/USER/ALL_LOB_SUBPARTITIONS : lob subpartition
information

- DBA/USER/ALL_IND_PARTITIONS, OBJECTS,
SEGMENTS : type index name lob information

CF. system generated partition name rule comparison
table partition : SYS_Pnn
table subpartition : SYS_SUBPnn
lob partition : SYS_LOB_Pnn
lob index partition : SYS_IL_Pnn
iot partition : SYS_IOT_TOP_Pnn
iot overflow partition : SYS_IOT_OVER_Pnn
CF. IOT system generated partition
name SYS_…_Pnn ‘P’
.

Index Organized Table

IOT

- Only range partition partition key
IOT primary key subset .
CF. , overflow segment table partition range .
(equipartitioned with the table partition)

- Secondary index partitioning local, global
index .

- General IOT partitioned IOT LOB or varray type
.

- partition maintenance partition command
. , LOB IOT
partition tablespace ‘move’ command
.

- Tablespace drop LOB partition
tablespace overflow
tablespace drop table partition
tablespace error return .
ORA-14404 : partitioned table contains partitions in a different
tablespace



- IOT partition

SQL> drop table emp_iot ;
Test
Table dropped.
IOT drop.
SQL> alter session set nls_date_format = 'DD-MON-YYYY' ;
Session altered.
Primary key name SQL> create table emp_of_iot (
. including 2 EMPNO NUMBER(4) not null,
hiredate
4 HIREDATE DATE,
overflow . 5 DEPTNO NUMBER(2),
Compress option 6 constraint xpkemp_of_iot primary key(empno, hiredate))
PK value 7 organization index including hiredate
. (prefix 8 compress 1
PK columns PK 9 overflow tablespace ts_emp6
columns –1 ) 10 partition by range(hiredate)
11 ( partition p_iot1 values less than ('01-JAN-1999') tablespace ts_emp1
Default nocompress
12 overflow tablespace ts_emp5,
compress default 13 partition p_iot2 values less than ('01-JAN-2000') tablespace ts_emp2) ;
PK column –1 . Table created.

SYS_IOT_OVER_ 2 from user_tab_partitions
overflow partition 3 where table_name in (select table_name from user_tables
. EMP_OF_IOT 4 where iot_name = 'EMP_OF_IOT')
tablespace 5 or table_name = 'EMP_OF_IOT' ;
IOT Table TABLE_NAME PARTITION_NAME TABLESPACE
. -------------------- -------------------- ----------
Index partition table SYS_IOT_OVER_42916 P_IOT2 TS_EMP6
SYS_IOT_OVER_42916 P_IOT1 TS_EMP5
partition
EMP_OF_IOT P_IOT1
. EMP_OF_IOT P_IOT2
P_iot1 partition
tablespace ts_emp5 SQL> select index_name, partition_name, tablespace_name
overflow
3 where index_name = 'XPKEMP_OF_IOT' ;
table level INDEX_NAME PARTITION_NAME TABLESPACE
ts_emp6 . --------------- -------------------- ----------
XPKEMP_OF_IOT P_IOT2 TS_EMP2
XPKEMP_OF_IOT P_IOT1 TS_EMP1

SQL> select segment_name, segment_type, tablespace_name
IOT table 3 where segment_name in (
storage 4 select table_name from user_tables
segment overflow 5 where iot_name = 'EMP_OF_IOT'
partition IOT partition 6 union
. 7 select index_name from user_ind_partitions
8 where index_name = 'XPKEMP_OF_IOT')
SEGMENT_NAME SEGMENT_TYPE TABLESPACE
-------------------- ------------------ ----------
SYS_IOT_OVER_42916 TABLE PARTITION TS_EMP6
XPKEMP_OF_IOT INDEX PARTITION TS_EMP2
XPKEMP_OF_IOT INDEX PARTITION TS_EMP1



SQL> create table emp_of_iot2 (
4 HIREDATE DATE,
5 DEPTNO NUMBER(2),
6 primary key(empno, hiredate))
7 organization index
8 overflow tablespace ts_emp3
9 partition by range(hiredate)
10 ( partition p_iot1 values less than ('01-JAN-1999') tablespace ts_emp1
PK name 11 overflow tablespace ts_emp4,
IOT . System 12 partition p_iot2 values less than ('01-JAN-2000') tablespace ts_emp2) ;
generated index name
SYS_IOT_TOP_Pnn Table created.
.
3 where index_name in ( select index_name
4 from user_indexes
5 where table_name = 'EMP_OF_IOT2') ;

INDEX_NAME PARTITION_NAME TABLESPACE
-------------------- -------------------- ----------
SYS_IOT_TOP_42934 P_IOT1 TS_EMP1
SYS_IOT_TOP_42934 P_IOT2 TS_EMP2

SQL> select segment_name, segment_type, tablespace_name
3 where segment_name in (
4 select table_name from user_tables
LOB 5 where iot_name = 'EMP_OF_IOT2'
6 union
IOT 7 select index_name from user_indexes
partition ts_emp2 8 where table_name = 'EMP_OF_IOT2') ;
drop tablespace
command SEGMENT_NAME SEGMENT_TYPE TABLESPACE
overflow partition -------------------- ------------------ ----------
ts_emp3, ts_emp4 SYS_IOT_OVER_42934 TABLE PARTITION TS_EMP4
SYS_IOT_TOP_42934 INDEX PARTITION TS_EMP1
drop .
SYS_IOT_TOP_42934 INDEX PARTITION TS_EMP2
IOT
partition ts_emp1, SQL> connect internal
ts_emp2 Connected

error return SQL> drop tablespace ts_emp2 including contents ;
. drop tablespace ts_emp2 including contents
*
ERROR at line 1:
ORA-14404: partitioned table contains partitions in a different tablespace

- IOT partition maintenance partition command
.



Object Type

Object type partitioning

- Oracle8i object table partitioning

- object type partitioning restriction :
nested partitioned table .
object type , REF, nested table or VARRAY partition
key .

- VARRAY LOB tablespace
. VARRAY
LOB
tablespace . , table partition
.

- .
SQL> create type emp_type as object (
2 empno number(4), ename varchar2(10), hiredate date) ;
3 /
Type created.

SQL> select type_name, typecode from user_types ;

TYPE_NAME TYPECODE
--------------- ----------
EMP_TYPE OBJECT
Type
SQL> desc emp_type
partitioned object table
Name Null? Type
. ------------ ------ -------------
EMPNO NUMBER(4)
ENAME VARCHAR2(10)
HIREDATE DATE

SQL> create table emp_obj_tp of emp_type
2 partition by range (empno)
3 (partition p_7000 values less than (8000) tablespace ts_emp1,
4 partition p_8000 values less than (9000) tablespace ts_emp2,
5 partition p_9000 values less than (maxvalue) tablespace ts_emp3) ;
Table created.

3 where table_name = 'EMP_OBJ_TP' ;

--------------- -------------------- ----------
EMP_OBJ_TP P_7000 TS_EMP1



SQL> select object_name, object_type, status
2 from user_objects
3 where object_id in ( select referenced_object_id
4 from public_dependency where object_id in ( select object_id
5 from user_objects where object_name = 'EMP_OBJ_TP')) ;

OBJECT_NAME OBJECT_TYPE STATUS
-------------------- -------------------- -------
Sys user view EMP_TYPE TYPE VALID
public_dependency
object table SQL> create type emp_va_php_type as varray(10) of number(15) ;
dependency object type 2 /
Type created.
.

TYPE_NAME TYPECODE
---------------- ------------
EMP_TYPE OBJECT
EMP_VA_PHP_TYPE COLLECTION

SQL> desc emp_va_php_type
emp_va_php_type VARRAY(10) OF NUMBER(15)

SQL> create type emp_va_profile as varray(1000) of varchar2(4000) ;
2 /
Type created.

Collection type varray
TYPE_NAME TYPECODE
partitioned object table ---------------- -----------
. type EMP_TYPE OBJECT
lob EMP_VA_PHP_TYPE COLLECTION
. EMP_VA_PROFILE COLLECTION

SQL> desc emp_va_profile
emp_va_profile VARRAY(1000) OF VARCHAR2(4000)

SQL> create table emp_obj_coll_tp (
2 empno number(4),
3 ename varchar2(10),
4 hiredate date,
5 emp_phone emp_va_php_type,
6 emp_profile emp_va_profile)
7 varray emp_profile store as lob (storage (initial 1m next 1m))
8 partition by range(empno)
9 (partition p_7000 values less than (8000) tablespace ts_emp1
10 varray emp_profile store as lob (storage (maxextents 1024)),
11 partition p_8000 values less than (9000) tablespace ts_emp3
12 varray emp_profile store as lob (storage (maxextents 2048)),
13 partition p_9000 values less than (maxvalue) tablespace ts_emp5) ;

Table created.



3 where table_name = 'EMP_OBJ_COLL_TP' ;

--------------- --------------- ----------
EMP_OBJ_COLL_TP P_8000 TS_EMP3

Table partition lob SQL> select table_name, partition_name, lob_partition_name, tablespace_name
partition tablespace
3 where table_name = 'EMP_OBJ_COLL_TP' ;
.
dependency TABLE_NAME PARTITION_NAME LOB_PARTITION_NAME
table object type TABLESPACE
. --------------- --------------- -------------------- ----------
EMP_OBJ_COLL_TP P_7000 SYS_LOB_P217 TS_EMP1

SQL> select object_name, object_type, status
2 from user_objects
3 where object_id in ( select referenced_object_id
4 from public_dependency where object_id in ( select object_id
5 from user_objects where object_name = 'EMP_OBJ_COLL_TP')) ;

OBJECT_NAME OBJECT_TYPE STATUS
-------------------- -------------------- -------
EMP_VA_PHP_TYPE TYPE VALID
EMP_VA_PROFILE TYPE VALID

- partition maintenance partition command
.

- Partitioned table tablespace drop
partitioned table tablespace
partition drop ORA-
14404 error return .



Parallel processing

Oracle8i parallel processing .

basic parallel processing Oracle8 New Features

.

Default degree of parallelism

- Oracle parallel degree
.

- SQL hint parallel degree
degree .

- table or index
parallelism .

- default degree
.
1. Number of CPU X parallel_threads_per_cpu(parameter)
2. OPS number of OPS instances in an MPP system
3. Number of partitions to be accessed

CF. , parallel_max_servers degree
.

Enable parallel DML/DDL

- parallel processing initial parameter parallel DML
enabling . session
level .

- SQL> alter session ENABLE/DISABLE/FORCE parallel
DML/DDL (parallel n) ;
CF. parallel query .
parallel query DML DDL .

- FORCE : parallel clause default
parallel degree .

- Oracle8 parallel DML commit or rollback
parallel
Oracle8i parallel DML object
parallel DML .
, parallel DML object read or
modify .



Check Data dictionary

- V$SESSION : PDML_ENABLED column
parallel enabling session level
.

- V$PQ_SYSSTAT : DML_INITATED query
Select * From V$PQ_SYSSTAT; initialized parallel DML

- V$PQ_SESSTAT : DML_PARALLELIZED query
Select * From V$PQ_SESSTAT; current session parallel

Review general restriction of parallel processing

- nonpartitioned table parallel update or delete
(error return serially )

- You cannot read or modify the same tables after modifying it
in parallel.

- You cannot read or modify the same tables after a serial direct
load INSERT.

- Parallel DML issues exclusive locks.
(no lock table or select for update statement can be in the same
transaction as the parallel DML statement)

- There is limited integrity constraint support.
1. no self-referential integrity
2. no delete cascade
3. no deferred integrity

- LOB or object type or IOT table clustered
table parallel DML .

- Replication . Trigger enable
Parallel DML statement serially
parallel DML trigger disable .
trigger replication
functionality .

- Distributed transaction . (serially
)

- Initial parameter ROW_LOCKING
. DML row locking
default always
default DML row level lock intent
table level lock .
intent table lock parallel DML
serial .



Parallel transaction recovery

- parallel DML rollback parallel .
user rollback recovery . ,
parallel DML rollback parallel
transaction rollback .

- , system or instance recovery large
transaction recovery
. PARALLEL_TRANSACTION_RECOVERY
dynamic initialization parallel
transaction recovery .

- PARALLEL_TRANSACTION_RECOVERY=
HIGH/LOW/FALSE (LOW is default)

- parameter LOW or HIGH SMON
recovery large transaction
parallel recovery candidate .
parallel transaction recovery SMON
coordinator parallel recovery processes slave
.

- parameter Oracle
BETA release version(8.1.3, 8.1.4)
.
CF. 13 Fast-Start parallel rollback .

- parallel transaction failed process recovery Oracle8
.
1. Single slave fail PMON rollback
slave rollback
.
2. Multiple slave fail PMON serially rollback
.
3. Coordinator process fail PMON coordinator
recover all slave processes parallel
rollback .
4. Process recovery time the forward work
2
.



Partition-Wise join in parallel

Prior Oracle8i, parallel 2 tables join query
step pieces scan and redistributed join
column CPU or memory(or temporary
tablespace) .

Oracle8i join key partition table
join (redistribution) skip
performance . ,
temporary storage memory
performance .(
setting Oracle
.)

parallel execution .

Full partition-wise join

- Equip partitioned tables join join columns
partitioned key table partition join
.

- tablespace equipartition local
disk .

Partial partition-wise join

- join column table partition
.( partition table or original table
)

- join column partition key table
redistribution .

CF. Oracle supports partition-wise joins on range, hash or
composite partitioned tables



Added parallel information

New parameter
- PARALLEL_AUTOMATIC_TUNING=TRUE/FALSE
parallel execution default degree Oracle
automatically control ’TRUE’
setting . , Default degree
Oracle .
parameter TRUE
1. A load balancing algorithm will be used which will evenly
distributed load across nodes in a multi-instance PQ
environment.
2. Oracle allocates parallel execution buffers from the large
pool. (FALSE allocates memory from the shared pool)

- PARALLEL_ADAPTIVE_MULTI_USER=TRUE/FALSE
parameter enables adaptive setting of the degree of
parallelism.
parallel_automatic_tuning ‘TRUE’
parameter ‘TRUE’ system load
degree of parallelism .

- PARALLEL_THREADS_PER_CPU=n
parameter default degree of parallelism
parallel adoptive algorithm
.

- Profile DEGREE_OF_PARALLELISM
database resource limit .
user maximum degree of parallelism
.( )

Dynamic performance views

- V$PX_PROCESS : running parallel execution session process
information

- V$PX_PROCESS_SYSSTAT : v$px_process parallel statistic

- V$PX_SESSION : running parallel execution session
information

- V$PX_SESSTAT : v$px_session and v$sesstat table joined
information for parallel execution



9. Universal Installer Migration and Upgrade

Universal Installer

Oracle8i New Features .

Universal Installer

- platform JAVA based installer
GUI install .

- Any customer can use the Software Packager to create
installable third-party components for oracle.

- Performing install from one location to several geographical
regions : Web Install Feature

Database Configuration Assistant

- Database configuration assistant

- Database creation(remote machine ) database
creation scripts easy parameter setting

- This is very useful for creating multiple database with the
same configuration.

- Run in silent mode
1. GUI OS command line
batch script
2. allows easy creation of multiple identical database

- Hardware detection and interview
1. File system, RAM
2. number of users,
3. database processing environment : OLTP, DSS..

Migration

Oracle8i New Features
Review .

What is changed ?

- migration upgrade
1. modify, add base tables and v$tables : data dictionary and
structures



2. modify datafile header blocks
3. change rollback segment format
4. control file recreation

Migration basic step
- Migration type
1. export / import
2. copy command
3. migration utility
3.1 command line migration utility
3.2 GUI data migration assistant

- Steps
1. prepare to migrate :
migration type , system resource
2. secure the required resources : resource
3. build a test plan : (OUTLN user or role drop)
migration test, minimal test, functional test,
integration test, performance test, volume/load stress test
4. preserve the source database :
database normal shutdown
cold backup
5. migrate the source database :
Oracle8i install
parameter or
migration ( migration type )
6. postmigration task :
shutdown normal
oracle8i cold backup
script (.sql file, recompile invalid objects )
startup

CF. Migration utility database .(
database rebuild)

CF. $ORACLE_HOME/rdbms/admin/utlrp.sql
upgrade invalid PL/SQL block modules
recompile .

Upgrade 8 to 8i
1. 8.0.4S --> upgrade 8.0.5 and upgrade 8.1

2. 8.0.3, 8.0.4, 8.0.5, 8.0.6 --> upgrade 8.1 directly



10. Tablespace Management

Extended Rowid

Oracle8 New Features review .

files rowid
( rowid files )

1022 tablespace 1022

partition object data object id

Rowid Type

- Restricted : BBBBBBBB.RRRR.FFFF
(Block Row File)
16 6 bytes

- Extended : OOOOOO.FFF.BBBBBB.SSS
(data Object relative File Block Slot)
64 10 bytes

DBMS_ROWID package check

Functions :

- ROWID_TO_EXTENDED : old rowid new rowid

- ROWID_TO_RESTRICTED : new rowid old rowid

- ROWID_VERIFY : old rowid
ROWID_TO_EXTENDED convert

- ROWID_TYPE : rowid old type new type

- ROWID_TO_ABSOLUTE_FNO : relative file number

- ROWID_OBJECT : object id return

- Others : ROWID_RELATIVE_FNO,
ROWID_BLOCK_NUMBER, ROWID_ROW_NUMBER

Procedures :

- ROWID_INFO : rowid

- ROWID_CREATE : rowid test



Locally-Managed tablespaces

tablespace dictionary managed tablespace
extent DD .
extent allocation DD update
system tablespace access rollback segment
.

Locally Managed (Oracle8i New Features)

- tablespace extent local datafile
bitmaps datafile .
DD access (system tablespace
access reduced) uniform size fragmentation
.

- Space management data dictionary datafile
bitmaps .
1. extent allocation resources DD access bitmaps
database-wide contention
2. SYSTEM tablespace
system tablespace
3. rollback segment

- SYSTEM tablespace tablespaces
.

Locally managed general tablespace

SQL> create tablespace ts_local_nis_d datafile
'/u01/local/data/nis_fin01.dbf' size 50M
extent management local uniform size 2M ;

Tablespace created.

- AUTOALLOCATE keyword UNIFORM
( autoallocate default
) storage initial size (initial extent
must be at least 64KB) other extents system
automatically size system-managed space
.

CF. Locally default storage .
, tablespace permanent/temporary
option temporary option .(
create temporary tablespace command
)

CF. UNIFORM size default 1M .



Locally managed temporary tablespace
SQL> create temporary tablespace ts_temp tempfile
’ /u01/local/temp/nis_temp01.dbf' size 100M
extent management local
uniform size 1M ;

Tablespace created.

- Oracle recommend locally management
tablespace temporary
tablespace .

- temporary tablespace
temporary option locally managed temporary tablespace
‘create temporary tablespace’
command . Command “tempfile”
option .

CF. tablespace locally managed
tablespace .

CF. Oracle 8.1.5 Manual System tablespace locally
managed tablespace dictionary managed
tablespace . , rollback segment
tablespace locally managed
.( , RBS extent system
tablespace )

CF. Oracle
“Locally managed system tablespaces are simply not yet possible.”



Data transportation

Transportable tablespaces

- OLTP staging area (or DW) DW Data
marts data point-in time recovery,
distributed relational data, archive historical data
.

- Export / Import tablespaces copy .

Procedure
1. tablespace read-only

2. export metadata

3. copy data files

4. transfer export dump file

5. import metadata

6. tablespace read-write (if you want)

Checking factors

1. Datafile O/S .

2. Oracle version 8.1 .

3. LOB table .

4. Block size and character set .

5. tablespace object tablespace
object move or exchange
partition command tablespace
.
(all partition tablespace )

6. Table nested tables or varray bitmap
index transport .

Checking availability

- Tablespace transportable
package .

- DBMS_TTS.TRANSPORT_SET_CHECK command
. (example )



Example

- tablespace transportation
.
O/S System_A

$ sqlplus internal
Connected.

SQL> create tablespace ts_trans datafile '/u01/trans/ts_trans01.dbf' size 10M ;
Tablespace created.

Connected.

SQL> create table emp_trans tablespace ts_trans
2 as select * from emp where rownum < 100 ;
Table created.

SQL> select table_name, tablespace_name from user_tables
2 where table_name = 'EMP_TRANS' ;
TABLE_NAME TABLESPACE_NAME
------------------------- ------------------------------
EMP_TRANS TS_TRANS

SQL> select count(*) from emp_trans ;
COUNT(*)
----------
Package check 99
tablespace
. SQL> connect internal
Connected.

SQL> exec dbms_tts.transport_set_check('TS_TRANS',TRUE) ;

SQL> select * from transport_set_violations ;
no rows selected

SQL> alter tablespace ts_trans read only ;
Tablespace altered.

SQL> exit

- transport_set_check tablespace
transport_set_violations view
row session
level .( session )

VIOLATIONS
--------------------------------------------------------------------------------
Partitioned table SCOTT.EMP_OBJ_COLL_TP is partially
contained in the transportable set: check table partitions by
querying sys.dba_tab_partitions ……..



- export system import

$ exp file=/u01/trans/exp_tbs_trans.dmp
> log=/u01/trans/exp_tbs_trans.log tablespaces=ts_trans
> transport_tablespace=y

Export: Release 8.1.6.2.0 - Production on Thu Aug 23 16:03:31 2001

(c) Copyright 1999 Oracle Corporation. All rights reserved.

Username: sys as sysdba
Password: change_on_install

Connected to: Oracle8i Enterprise Edition Release 8.1.6.2.0 - Production
……..
. exporting cluster definitions
. exporting table definitions
. . exporting table EMP_TRANS
…….
Export terminated successfully without warnings.

O/S System_B

$ ftp system_A
Connected to system_A.
….
ftp> bi
200 Type set to I.
ftp> get exp_tbs_trans.dmp
…
226 Transfer complete.
ftp> get ts_trans01.dbf
…
226 Transfer complete.
ftp> quit
Datafile export dump 221 Goodbye.
file ftp
import . $ imp file=exp_tbs_trans.dmp
> log=/u01/trans/imp_tbs_trans.log transport_tablespace=y
> datafiles=/u01/trans/ts_trans01.dbf

Import: Release 8.1.6.2.0 - Production on Thu Aug 23 16:17:43 2001

(c) Copyright 1999 Oracle Corporation. All rights reserved.

Username: sys as sysdba
Password: change_on_install

Connected to: Oracle8i Enterprise Edition Release 8.1.6.2.0 – Production
…
. importing SYS's objects into SYS
. importing SCOTT's objects into SCOTT
. . importing table "EMP_TRANS"
Import terminated successfully without warnings.



- import .
$ sqlplus internal

SQL> select tablespace_name, file_name from dba_data_files
2 where tablespace_name = 'TS_TRANS' ;

TABLESPACE FILE_NAME
---------- ------------------------------
TS_TRANS /u01/trans/ts_trans01.dbf
Import System B SQL> conn scott/tiger
tablespace ts_trans Connected.
creation
tablespace object SQL> select table_name, tablespace_name from user_tables
data System A 2 where table_name = 'EMP_TRANS' ;
System B
TABLE_NAME TABLESPACE
.
--------------------------- ----------
EMP_TRANS TS_TRANS

SQL> select count(*) from emp_trans ;

COUNT(*)
----------
99

SQL> connect internal
Connected.
SQL> alter tablespace ts_trans read write ;

Tablespace altered.

- System B emp_trans table
write tablespace
.

CF. import fromuser, touser options schema that
owns the object in transported tablespace another schema
.

Read-Only tablespace

- read only tablespace active
transaction
.

- , current transaction waiting read-
only .

CF. shutdown normal .



11. Database resource manager

Resource management

DBA resource management resource control
resource .

DBA resource plan plan
consumer group, resource allocation method, resource plan
directives resource
.

Oracle user resource
Oracle8i New Features .

EX. daytime resource plan

Resource consumer group Allocation method parameters
OLTP CPU=70%
Degree of parallelism = 0
DW CPU=30%
Degree of parallelism = 18

Resource management factors

- resource consumer group : plan resource
user group

- resource plan : CPU time parallel query resources
. resource plan resource plan
.
DBA DB shutdown plan switch .
EX. daytime plan nighttime plan switch
database restart .

- resource plan directives :
1. assign consumer groups or subplans to resource plans
2. allocate resources among consumer groups or subplans

Resource allocation method

- CPU usage :
1. level 1 to level 8
2. percentages specifying how to partition CPU at each level

- Degree of parallelism : maximum degree of parallelism



Resource consumer groups

- session group active .

- Users consumer groups .

- Default group login .

- Session level user or DBA can switch consumer group

Steps in Database Resource Management

Steps for database resource management creation
1. create resource plan

2. create resource consumer groups

3. create resource plan directives

4. assign users to consumer groups

5. specify plan to be used by an instance

Packages for DB resource manager
- DBMS_RESOURCE_MANAGER :
1. define plans and classes, plan entries
2. define the default class for a user
3. alter the class for currently connected user

- DBMS_RESOURCE_MANAGER_PRIVS
1. grant and revoke classes to users
2. switch the class for a session

Create resource plan
- DBMS_RESOURCE_MANAGER.CREATE_PENDING_AREA ;
pending area :
1. store new resource manager plans, consumer groups and
resource plan directives
2. changes before they are committed.

- DBMS_RESOURCE_MANAGER.CREATE_PLAN (
’daytime’, ‘from 15 to 19’) ;
==> ‘plan name’, ‘comment for this plan’
CF. database creation default SYSTEM_PLAN
plan .



- Privileges needed to manage resources
1. administer_resource_manager :
package for dbms_resource_manager
2. grant_system_privilege and
revoke_system_privilege .
EX. DBMS_RESOURCE_MANAGER_PRIVS.
GRANT_SYSTEM_PRIVILEGE(‘scott’, admin_option =>
TRUE) ;
==> ‘username’, with ADMIN_OPTION

Creating resource consumer groups
- DBMS_RESOURCE_MANAGER.
CREATE_CONSUMER_GROUP (‘oltp’, ‘online job working’) ;
==> ‘group name’, ‘comment for group’

- Database group
1. SYS_GROUP : SYSTEM_PLAN high priority
for ‘SYS’ and ‘SYSTEM’ users
2. LOW_GROUP : low priority group in SYSTEM_PLAN
3. DEFAULT_CONSUMER_GROUP :
assigned to all users(PUBLIC) by default
4. OTHER_GROUPS : a part of every plan
group plan other_groups directives
. plan each top-plan
other_groups directive .
( example .)

Create resource plan directives

- DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE
(‘daytime’, ‘olpt’, ‘rule for oltp job’, cpu_p1 => 70,
parallel_degree_limit_p1 => 18) ;
==> ‘plan name’, ‘group name’, ‘comment for directive’, ‘CPU
level 1’, ‘parallel degree’

- cpu_p2 multi level CPU
?

- , level1 OLTP DW 70:30
level2 DSS Web 40:60
. Level2 CPU
.

- system load cpu 80%
. level2 20%
level2 cpu
. , CPU level %
. Priority level %
.



Confirming resource management

- resource factors pending area
validate and confirm resource
factors .

- : DBMS_RESOURCE_MANAGER.
VALIDATE_PENDING_AREA ;
1. all consumer groups referred to in the plans created or
changed exist
2. plans have directives that reference other plans or consumer
groups
3. plans do not reference themselves either directly or
indirectly
4. the sum of CPU percentages defined for a specific level
within a plan do not exceed 100
5. a plan that is currently used is not deleted

- implicitly validate and commit :
DBMS_RESOURCE_MANAGER.
SUBMIT_PENDING_AREA ;

Assign users to consumer groups
- assign user or role to consumer group
( package user or role consumer group
)
DBMS_RESOURCE_MANAGER_PRIVS.
GRANT_SWITCH_CONSUMER_GROUP
(‘scott’, ‘oltp’, FALSE)
==> ‘user or role’, ‘group name’, with GRANT option

- scott user consumer group .
user group
session active group . ,
application
DBA Initial
consumer group for users ( group assign
user default group)
session level group .

- scott user oltp group
.
SET_INITIAL_CONSUMER_GROUP(‘scott’,’oltp’) ;

- session level consumer group
.



Changing consumer group within a session

- group session
user .

- user or application level (PL/SQL block )
DBMS_SESSION.SWITCH_CURRENT_CONSUMER_GROUP
(‘oltp’, old_group) ;
==> ‘new group’, old group (output variable)

- DBA
SWITCH_CONSUMER_GROUP_FOR_SESS
(453, 23412, ‘oltp’) ;
==> sid, serial#, ‘group name’

- DBA all sessions for a user
SWITCH_CONSUMER_GROUP_FOR_USER
(‘scott’, ‘oltp’) ;
==> ‘username’, ‘group name’

How to set resource plan for an instance

- Specify plan for initial parameter :
resource_manager_plan=daytime

- Change plan without shutdown and restart :
alter system set resource_manager_plan=nighttime

- Initial parameter setting plan
database open error return .
resource plan parameter
shutdown and restart . parameter
default value null .

Data dictionary

>> RESOURCE MANAGER INFORMATION

- DBA_RSRC_PLANS : resource plans and status

- DBA_RSRC_PLAN_DIRECTIVES : resource plan directives
and status

- DBA_RSRC_CONSUMER_GROUPS : consumer groups and
status

- DBA_RSRC_CONSUMER_GROUP_PRIVS : users granted
consumer groups

- DBA_USERS : new column ‘initial_rsrc_consumer_group’



>> CURRENT SETTINGS

- V$SESSION : new column ‘resource_consumer_group’

- V$RSRC_PLAN : showing the active resource plan

- V$RSRC_CONSUMER_GROUP : showing the number of
active sessions by consumer group and wait/yield statistics

Example

- plan,
group, directive resource management .
$ sqlplus internal
SQL> select plan, group_or_subplan, type from dba_rsrc_plan_directives ;
PLAN GROUP_OR_SUBPLAN TYPE
--------------- ------------------------------ --------------
SYSTEM_PLAN SYS_GROUP CONSUMER_GROUP
SYSTEM_PLAN OTHER_GROUPS CONSUMER_GROUP
SYSTEM_PLAN LOW_GROUP CONSUMER_GROUP

SQL> select * from dba_rsrc_consumer_group_privs ;
GRANTEE GRANTED_GROUP GRA INI
---------- ------------------------------ --- ---
PUBLIC DEFAULT_CONSUMER_GROUP YES NO
PUBLIC LOW_GROUP NO NO
SYSTEM SYS_GROUP NO YES

SQL> exec dbms_resource_manager.create_pending_area ;

6 SQL> exec dbms_resource_manager.create_plan('daytime', 'oltp from 15 to 19') ;
plan set . PL/SQL procedure successfully completed.
Pending area --> Plan -->
Group --> Directive --> SQL> exec dbms_resource_manager.create_consumer_group('oltp', 'online job
working') ;
Check(validate) -->
Commit(submit) ;
plan SQL> exec dbms_resource_manager.create_plan_directive('daytime', 'oltp', 'rule
other_groups for oltp job', cpu_p1 => 70, parallel_degree_limit_p1 => 18 ) ;
. PL/SQL procedure successfully completed.

SQL> exec dbms_resource_manager.validate_pending_area ;
BEGIN dbms_resource_manager.validate_pending_area ; END;
*
ERROR at line 1:
ORA-29382: validation of pending area failed
ORA-29377: consumer group OTHER_GROUPS is not part of top-plan
DAYTIME
ORA-06512: at "SYS.DBMS_RMIN", line 55
ORA-06512: at "SYS.DBMS_RESOURCE_MANAGER", line 254
ORA-06512: at line 1



SQL> exec dbms_resource_manager.create_plan_directive('daytime',
'other_groups', 'other groups for daytime users', cpu_p1 => 0,
parallel_degree_limit_p1 => 0 ) ;

SQL> exec dbms_resource_manager.validate_pending_area ;
SQL> exec dbms_resource_manager.submit_pending_area ;
other_groups PL/SQL procedure successfully completed.
commit
. SQL> select plan, group_or_subplan, type from dba_rsrc_plan_directives ;
OLTP group PLAN GROUP_OR_SUBPLAN TYPE
other_groups . --------------- ------------------------------ --------------
DAYTIME OTHER_GROUPS CONSUMER_GROUP
SYSTEM_PLAN SYS_GROUP CONSUMER_GROUP
SYSTEM_PLAN OTHER_GROUPS CONSUMER_GROUP
SYSTEM_PLAN LOW_GROUP CONSUMER_GROUP
DAYTIME OLTP CONSUMER_GROUP
Grant scott user
SQL> exec dbms_resource_manager_privs.grant_switch_consumer_group('scott',
consumer group 'oltp', FALSE) ;
.( PL/SQL procedure successfully completed.

.) SQL> exec dbms_resource_manager.set_initial_consumer_group('scott', 'oltp') ;

SQL> select username, initial_rsrc_consumer_group from dba_users
2 where username = 'SCOTT' ;
USERNAME INITIAL_RSRC_CONSUMER_GROUP
-----------------------------------------
SCOTT OLTP Scott user
session B .
session B $ sqlplus scott/tiger
Internal
Connected.
session A .
session A SQL> select sid, serial#, username, resource_consumer_group
2 from v$session where username= 'SCOTT' ;
SID SERIAL# USERNAME RESOURCE_C
---------- ---------- --------- ----------
11 18058 SCOTT OLTP
session B SQL> set serveroutput on
SQL> declare
2 v_old_group varchar2(30) ;
3 begin
4 dbms_session.switch_current_consumer_group('default_consumer_group',
v_old_group, false) ;
5 dbms_output.put_line(v_old_group) ;
Session group 6* end;
. Resource 7/
manager off declare
. *
ERROR at line 1:
ORA-29373: resource manager is not on
ORA-06512: at "SYS.DBMS_SESSION", line 74



- Session level group dbms_session package
error .

- initial parameter resource_manager_plan
resource manager off .
alter system command plan
parameter resource manager on
.
SQL> select * from v$rsrc_plan ;
session B
no rows selected

session A SQL> SQL> alter system set resource_manager_plan = 'daytime' ;
System altered.

session B SQL> select * from v$rsrc_plan ;

Resource manager NAME
turned on --------------------------------
daytime

SQL> declare New Group Name
2 v_old_group varchar2(30) ;
3 begin
4 dbms_session.switch_current_consumer_group('default_consumer_group',
Current session
v_old_group, false) ;
consumer group 5 dbms_output.put_line(v_old_group) ;
. 6* end;
7/ session
group .
OLTP


SQL> select sid, serial#, username, resource_consumer_group
session A 2 from v$session where username= 'SCOTT' ;
SID SERIAL# USERNAME RESOURCE_C
---------- ---------- --------- ----------
group
11 18058 SCOTT DEFAULT_CONSUMER_GROUP

-
.

delete_plan procedure 1. alter system set resource_manager_plan = ‘’ ;
subplan or 2. dbms_resource_manager_privs.revoke_switch_consumer_group
consumer groups ('scott', 'oltp') ;
3. dbms_resource_manager.create_pending_area ;
.
4. dbms_resource_manager.delete_plan_cascade('daytime') ;
delete_plan_cascade
5. dbms_resource_manager.submit_pending_area ;
drop . (plan, directives,
CF. plan drop package
subplan, consumer group)
update_plan, update_consumer_group, update_plan_directive,
delete_plan plan
delete_consumer_group, delete_plan_directive .
directives delete .



12. Manageability Features

Database limits

Oracle8 and Oracle8i database limits

- Oracle8 New Features review

Columns per table 1,000

Columns per index 32

CAHR, NCHAR 2,000 bytes

VARCHAR2, NVARCHAR2 4,000 bytes

CLOB, BLOB, BFILE 4GB

Tablespaces per database 65,533

Datafiles per tablespace 1,022

Datafiles per database 65,533

- only one LONG column per table

- index define column total size database
block 40% bitmapped index
index 30 .

- Tablespace limit 65,533 tablespace 1
datafile .

- Tablespace datafile O/S 1,022
.

- Datafile initial parameter DB_FILES .



Move Table

Relocating and Reorganizing a Table
SQL> alter table emp move tablespace ts_new_emp ;

- builds new segments and drops old segments

- used to move table to a different tablespace or reorganize
extents

- limits
1. non-partitioned table only (partitioned table error)
2. move only select available and no DML
3. index, constraints and grant data
index status UNUSABLE
.
4. new segment export and import
table 2 space .

Remove Column

Marking a column as unused
SQL> alter table table_name set unused column column_name ;

- table column command users
column data
disk space
.

- unused column physically
remove .

- user unused column column
.

- Unused column table export drop and
import table unused column
.

SQL> select * from dba_unused_col_tabs ;
unused column information .



Removing a column from table
SQL> alter table table_name drop column column_name [cascade
constraints] ;

- command unused mark
column drop .( space release) , large
table command
.

- column ‘UNUSED’ mark
column command drop .

- cascade constraints : drop column all
constraints drop .

SQL> alter table table_name drop unused columns ;

- table removes all currently marked as unused

CF. dropped column object , views, triggers and
stored procedure units status invalidate .

CF. dropped column index always drop
automatically

Minimizing rollback segment usage
SQL> alter table table_name drop column columns_name
[ checkpoint 1000 ] ;

- DDL rollback segment checkpoint
option row commit
.

- 1000 row checkpoint .

- Checkpoint table drop
rollback space .

- instance failure
table INVALID state .

- command .
SQL> alter table table_name drop columns continue ;



Example

- emp table test emp_test
.
SQL> desc emp_test
Name Null? Type
------------ -------- -----------
EMPNO NOT NULL NUMBER(4)
ENAME VARCHAR2(10)
SAL NUMBER(7,2)
COMM NUMBER(7,2)
DEPTNO NUMBER(2)

SQL> alter table emp_test drop column deptno
2 cascade constraints checkpoint 1000 ;

Table altered.
constraints drop
SQL> desc emp_test
1000 row Name Null? Type
commit column ------------ -------- -----------
ENAME VARCHAR2(10)
SAL NUMBER(7,2)
COMM NUMBER(7,2)

SQL> alter table emp_test set unused column comm ;

Table altered.

SQL> select * from user_unused_col_tabs ;

TABLE_NAME COUNT
unused -----------------------
EMP_TEST 1
unused columns
SQL> desc emp_test
Name Null? Type
------------ -------- -----------
ENAME VARCHAR2(10)
SAL NUMBER(7,2)

SQL> alter table emp_test drop unused columns ;

Table altered.

SQL> select * from user_unused_col_tabs ;

no rows selected



Temporary tables

Application global temporary table

- temporary table data (DML ) session
visible truncate command . , table
transaction or session termination data
drop .

CF. , session instance crash data
drop . (temporary )

- transaction or session data . (index
index Index data )

- Table definition definition all session
. (definition export/import )

- Session sort space data
user’s temporary tablespace .

- Permanent tables view temporary
table triggers .

- Recovery data redo entries
DML rollback segment .

Creating and Monitoring with sample
SQL> create global temporary table table_name
(empno number, ename varchar2(10), sal number)
on commit delete rows ;

- on commit delete rows : transaction
transaction session
temporary table data .

- on commit preserve rows : session (transaction
session )

SQL> select table_name, tablespace_name, temporary, duration
from dba_tables
where table_name = ‘TABLE_NAME’ ;

- dba_tables table temporary
transaction session
.



Example

- scott user

SQL> create global temporary table emp_temp_sess
2 (empno number, ename varchar2(10), sal number)
3 on commit preserve rows ;
Table created.
SQL> create global temporary table emp_temp_trans
2 (empno number, ename varchar2(10), sal number)
3 on commit delete rows ;
type global Table created.
temporary table
system user SQL> connect system/manager
Connected.
type temporary table
SQL> select table_name, tablespace_name , temporary, duration
. 2 from dba_tables
3 where table_name like '%TEMP%' and owner = 'SCOTT' ;

TABLE_NAME TABLESPACE T DURATION
-------------------- ---------- - ---------------
EMP_TEMP_SESS Y SYS$SESSION
EMP_TEMP_TRANS Y SYS$TRANSACTION

SQL> connect scott/tiger
Connected.
SQL> insert into emp_temp_sess values ( 1000, 'KS JANG', 100000) ;
1 row created.
Scott user table SQL> insert into emp_temp_trans values ( 1000, 'KS JANG', 100000) ;
data insert data 1 row created.
.
SQL> select * from emp_temp_sess ;
EMPNO ENAME SAL
---------- ---------- ----------
1000 KS JANG 100000
SQL> select * from emp_temp_trans ;
EMPNO ENAME SAL
---------- ---------- ----------
1000 KS JANG 100000

SQL> commit ;
Commit complete.
Commit select
emp_temp_trans table SQL> select * from emp_temp_sess ;
EMPNO ENAME SAL
transaction data
---------- ---------- ----------
1000 KS JANG 100000
emp_temp_sess session
data SQL> select * from emp_temp_trans ;
. no rows selected

SQL> connect scott/tiger
Connected.
SQL> select * from emp_temp_sess ;
no rows selected

SQL> select * from emp_temp_trans ;
no rows selected



SQL*Loader

Enhancements

- TERMINATED BY clause 1 character strings
. (multiple delimited field available)

- New variable length field : VARCHARC, VARRAWC
EX. VARCHARC(3,200) : ‘ 3 value
. 4 character
character read 200
’ .
CF. VARRAWC data RAW type .

- record size 64K .
However, a logical record must fit entirely in client memory.

EX. enhancements .
LOAD DATA
INFILE ‘emp_data.dat’
Record RECSEPARATOR ‘;’
Field FIELDS TERMINATED BY ‘::’
(emp_name, location NULLIF
Multiple
(emp_name=BLANKS AND location=BLANKS)
New field type
comment VARCHARC(3) )

Kate::Seoul::015Shi is actress;Bill::Pusan::024He is the best actor



Monitoring Long-Running Operations

V$SESSION_LONGOPS

- dynamic performance view monitoring

- monitoring operation
.
1. arching
2. RMAN backup and restore
3. parallel query
4. recovery
5. full table scan
6. sort operation
7. analyze using DBMS_STAT
8. hash cluster creation

EX. monitoring .
SQL> connect system/manager
Connected.
SQL> select sid, serial#, opname,
2 to_char(start_time, 'HH24:MI:SS') startt,
3 (sofar/totalwork)*100 "complete(%)"
4 from v$session_longops
5 where totalwork > 0 ;

SID SERIAL# OPNAME STARTT complete(%)
13 238 TABLE SCAN 11:00:09 49.8434
45 342 ANALYZE 11:03:29 21.3433

- DBMS_APPLICATION_INFO package
SET_SESSION_LONGOPS procedure
application view populate .

- PL/SQL Block
dbms_application_info.set_session_longops(v_rindex, v_slno) ;
session v$session_longops
row . , v_rindex, v_slno IN/OUT variable
declare PL/SQL Block
.

CF. package application operating
information v$session_longops . ,
row creation . session
session termination delete .

CF. package application
monitoring .



13. Availability and Recoverability

Archive Enhancements

Multiple archive logs

- Oracle7 Oracle archive log 1
destination Oracle8
log_archive_duplex_dest
. , parameter remote instance
destination .

- Oracle8i Oracle8 archive dest
parameter log_archive_dest_n
(n 1~5 ) 5 archive dest
remote instance .
remote backup system standby database
. , remote instance remote
instance 1 log_archive_dest_n .

- log_archive_dest log_archive_dest_n
. log_archive_dest_n
Oracle Enterprise Edition install
Oracle8.1.5 Reference Manual
log_archive_duplex_dest .

CF. ( ) Oracle Version 8.1.6.2 “Enterprise Edition”
test Enterprise Edition
log_archive_duplex_dest .

-
1. 2 archive dest ,
2. standby database
3. remote archive destination
log_archive_dest_n .

CF. standby database 2 local
archive destination 1 remote archive destination
.

- Oracle8i remote archive destination
log_archive_dest_n session level
dynamic parameter



Multiple archive log processes

- Oracle8i multiple archive process
. log_archive_start=true setting
log_archiv_max_processes
archive process start .

CF. instance start alter system command
. (10 )
alter system set log_archive_max_processes=2 ;

- archiving throughput (multiple archive processes
archive ) Oracle
version manual archives
performance .

Multiple archive Options

- , standby database
1 local archive dest 1 backup local archive dest
.

log_archive_dest_1 =
“location=/a01/oracle1/arch/arch mandatory reopen”
“location=/a02/oracle2/arch/arch optional”
“service=standby_db2 mandatory reopen=500”

- dest #1 original archive dest
dest #2 archive backup dest dest #3 standby
database .

- dest #1 dest #3 mandatory
archive .
reopen archive
default 300 dest #3
500 . second .

- dest #2 optional redo log
reused .

- dest #3 tnsnames.ora service name standby
database archive .
standby database ‘RFS’ process primary database
.

CF. as mandatory / optional optional default
value .



When is the online redo log reused?

- online redo log reused
.

1. archive all mandatory locations

2. log_archive_min_succeed_dest local location
archived (all local mandatory archive
)

- standby database archive
destination mandatory archive
local archived
log_archive_min_succeed_dest archive
online redo log reuse
.

- log_archive_min_succeed_dest total archive
destination local mandatory archive
destination . ,
.

- 3 local archive destination
mandatory 2 standby database archive mandatory 1
optional 2 online redo log
reuse .

- 3 mandatory archive
2 local archive destination 1
optional .

CF. , log_archive_dest log_archive_duplex_dest
parameter 1 2 .

- log_archive_min_succeed_dest parameter session or system
level .
alter session set log_archive_min_succeed_dest=2 ;

alter system set log_archive_min_succeed_dest=3 ;
CF. log_archive_dest log_archive_duplex_dest
alter session command .

- parameter log_archive_dest_state_n
_n archive destination disable .
default ‘ENABLE’ session or system level
. ‘DEFER’ destination archive
‘ENABLE’ archive .



- , ‘DEFER’ ‘ENABLE’ archive
destination manual copy .
alter system set log_archive_dest_state_3 = defer ;
alter system set log_archive_dest_state_1 = enable ;
(log_archive_dest_n destination
log_archive_dest_state_n matching .

CF. Anyway, log_archive_min_succeed_dest
‘ENABLE’ log_archive_dest_state_n .

Related data dictionary

- V$ARCHIVE_DEST : archive destination
BINDING, STATUS, TARGET, FAIL_SEQUENCE……

- V$ARCHIVE_PROCESSES : archive process
PROCESS, STATUS, LOG_SEQUENCE, STATE



Standby database

Automatic Archival

- primary site archive log automatically shipped to
standby site archive manual copy
.

- Primary site archive ‘RFS’
(Remote File Server) process .
process standby database instance start
remote node file save control file
update .

- If multiple ARCH processes are used at the primary site, each
ARCH process spawns and communicates with its own RFS
process.

Sustained Recovery

- Optionally, RFS primary site log
standby database . (Sustained Recovery)
primary failure standby open
.

- Primary archive destination
standby_archive_dest=/primary_site/a01/arch…..

- Sustained recovery command
1. mount .
2. alter database recover managed standby database ;
[timeout 60] ==> primary log
. (minutes )

Read only database

- select database open .
standby database production .

- Read only database
1. query .
2. order by sort , locally managed
tablespace sort .
3. datafile online or offline .

- Standby read only open sustained recovery
1. alter database recover managed standby database cancel ;
2. alter database open read only ;
3. alter database recover managed standby database ;
(active user sessions )



Redo log Enhancements

LogMiner : PL/SQL interface redo log files
read, analyzed, interpreted using
standard SQL .

Analyzing redo log files
- database changes tracking without auditing overhead
time or SCN base , point-in-time recovery
.

- Performing granular logical recovery
application transaction undoing
specific changes .

- Tuning and capacity planning
data growth patterns usage patterns
.

LogMiner (install ‘SYS’ user )

- Oracle8 redo log .
1. same platform
2. same character set

- Log file dictionary file
.

- dbms_logmnr_d.build(‘file_name’,’file_path’) ;

CF. Oracle8 log analyzing , Oracle8i
$ORACLE_HOME/rdbms/admin/dbmslogmnrd.sql
Oracle8 copy ‘SYS’ user .

CF. Oracle 8.1.6.2 dbmslmd.sql
. (dbmslogmnrd.sql )

- file_path initial parameter UTL_FILE_DIR
.

- procedure data dictionary information
file object
. , object
.

- procedure log table index, data
scripts . file
target database



database
log .

- dbms_lomnr_d.build log
object
name data hexa value
.

- Log file log
.

- dbms_logmnr.add_logfile(‘/a01/oracle/arch/arch_log1’,
dbms_logmnr.NEW) ;
LOG

dbms_logmnr.ADDFILE) ;
LOG LOG

dbms_logmnr.REMOVEFILE) ;
LOG LOG

CF. test log file
.

- log file log initiate
.

- Log SCN Number
.
(dbms_logmnr_d.build file dictfilename
path )

- dbms_logmnr.start_logmnr(
dictfilename=>’/app/oracle/logmnr/ logmnr.ora’, -
start_scn => 350128, -
end_scn => 352286) ;

dbms_logmnr.start_logmnr(
dictfilename => ‘/app/oracle/logmnr/logmnr.ora’, -
starttime => to_date(‘31/05/2001:11:35’,
‘DD/MM/YYYY:HH24:MI’), -
endtime => to_date(‘31/05/2001:14:18’,
‘DD/MM/YYYY:HH24:MI’) ) ;
CF. line ‘-‘ .
CF. SCN or time range addfile log file
procedure error
.” time or SCN range not fully contained in the listed
logfiles” .
CF. , dictfilename log file



. starttime endtime default
value 1988 1 1 2988 1 1
.

- LogMiner start
V$LOGMNR_CONTENTS View .

- ‘AKCDFILE’ Table insert, delete
2 commit
2 transaction
.
SQL> select scn, timestamp, username, sql_redo, sql_undo from v$logmnr_contents where seg_name = 'AKCDFILE' ;

SCN TIMESTAMP USERNAME SQL_REDO
350128 20010531 11:35:21 AKMANAGER insert into AKMANAGER.AKCDFILE (FILE_TP,FILE_RUN)
values ('AVI','audio file');
SQL_UNDO
delete from AKMANAGER.AKCDFILE where FILE_TP = 'AVI' and
FILE_RUN = 'audio file' and ROWID = 'AAAA2PAAHAAAACFAAE';

350138 20010531 11:36:14 AKMANAGER delete from AKMANAGER.AKCDFILE where FILE_TP = 'AVI' and
FILE_RUN = 'audio file' and ROWID = 'AAAA2PAAHAAAACFAAE';
SQL_UNDO
insert into AKMANAGER.AKCDFILE (FILE_TP,FILE_RUN) values
('AVI','audio file');

CF. SQL REDO user SQL UNDO
. , SQL_REDO
SQL_UNDO recovery
.

- log
end_logmnr resource release .

- dbms_logmnr.end_logmnr ;

- log . , log
session .

- dbms_logmnr.procedure
session PGA
dbms_logmnr.end_logmnr PGA
.

- Other information :
1. V$LOGMNR_DICTIONARY
- tiemstamp, filename
2. V$LOGMNR_LOGS
- log id, log filename, low-high scn, log-high time
3. V$LOGMNR_PARAMETERS : arguments
- start-end date, start-end scn, info, status



Limitation

- PGA session .

- 1 row per record (1 row
sel_undo row sql undo
rowid ) ,
row return .

- DDL oracle , data dictionary
DML .

- Chained rows DML reconstruct
.

- Data dictionary file clustered tables
changes hexa values . ,
dictionary file hexa values
.

CF. DDL sample ‘xx’ table
create and drop log
.
- drop table xx ….
sql_redo : delete from SYS.OBJ$ where OBJ# = 3501 and DATAOBJ# = 3501 and OWNER# = 0 and NAME
= 'XX' and NAMESPACE = 1 and SUBNAME IS NULL and TYPE# = 2 and CTIME = TO_DATE('31-MAY-
2001 15:36:10', 'DD-MON-YYYY HH24:MI:SS') and MTIME = TO_DATE('31-MAY……………….
sql_undo : insert into SYS.OBJ$
(OBJ#,DATAOBJ#,OWNER#,NAME,NAMESPACE,SUBNAME,TYPE#,CTIME,MTIME,STIME,STATUS,
REMOTEOWNER,LINKNAME,FLAGS,OID$,SPARE1) values (3501,3501,0,'XX',1,NULL,2,
TO_DATE('31-MAY-2001 15:36:10', 'DD-MON-YYYY HH24:MI:SS'),TO_DATE('31- ……………..



Fast_Start Recovery

Instance Recovery

- Prior to oracle8i instance recovery time
. failure transaction load
.

- fast_start_io_target parameter : instance recovery
(read and write) maximum number of
datafile blocks .( parameter target
redo blocks , redo generation )
system level dynamic parameter instance running
alter system command .

- Recovery time datafile I/O instance
average I/O time .

Fast-Start Checkpointing

- crash or instance recover recovery position.
buffer cache earliest buffer to be dirtied gets written first
(oldest dirty block buffer) roll
forward DB blocks
. , feature media recovery .

- If fast_start_io_target = 10000 10000 blocks
checkpoint position .

- parameter recovery
performance checkpoint
transaction overhead .

Factors affecting checkpointing
- A : fast_start_io_target parameter value

- B : 90% of size of smallest redo log

- C : end of the log_checkpoint_timeout seconds ago

- D : log_checkpoint_interval blocks from the end

DBWn 4 redo log
point satisfy all the criteria defined.
Redo Log blocks end of the redo

D B C A
CF. redo ‘A’ point .



Factors influencing time to recover

- estimated recovery time
. recovery time
.

- Checkpoint target changed only at specific time intervals
(maybe time recovery?)

- Additional recovery activities such as reading logs and lock
remastering .
(typically recovery activities 5% )

- Recovery time may be faster if parallel recovery is used

Define and monitor bounded recover time

- Users V$FILESTAT AVGIOTIM column
FAST_START_IO_TARGET .
( column TIMED_STATISTICS initial parameter
TRUE 1/100 file I/O
)

- V$INSTANCE_RECOVERY impact of parameters
check .
1. RECOVERY_ESTIMATED_IOS : fast-start checkpoint
.
2. ACTUAL_READ_BLKS : recovery current actual
redo blocks
3. TARGET_READ_BLKS : 4 minimum
value recovery current target
redo blocks . the number of
redo blocks.
4. LOG_FILE_SIZE_REDO_BLKS : checkpoint complete
log switch
5. LOG_CHKPT_TIMEOUT_REDO_BLKS :
log_checkpoint_timeout parameter
6. LOG_CHKPT_INTERVAL_REDO_BLKS :
log_checkpoint_interval parameter
7. FAST_START_IO_TARGET_REDO_BLKS :
fast_start_io_target parameter

CF. DB_BLOCK_MAX_DIRTY_TARGET
FAST_START_IO_TARGET 0 DBMDT
trigger checkpoint .
FAST_START_IO_TARGET parameter
DBMDT override .

CF. FSIT recovery blocks
DBWn dirty buffers write
.



Fast-Start parallel rollback

Rolling back

- Oracle fast start Oracle7 warm start , deferred
rollback .

- Rolling back
1. SMON periodically scans all rollback segments and
rolls back aborted transactions as needed
2. user process dead transaction held row
lock recovers transaction and continues processing.

- Oracle8i instance recovery SMON query slave
query slave process 1
rollback parallel rollback
.

- Parallel transaction system crash recovery
.
10 10 parallel instance
crash rollback 100
.

- SMON parallel query slaves parallel
rollback operation .

- , SMON dead transaction large number of
rollback blocks generate (rollback
blocks 100 ) .

Controlling fast-start parallel rollback

- Oracle 8.1.4 Beta Release
PARALLEL_TRANSACTION_RECOVERY parameter
parameter .
(8 parallel processing parallel transaction rollback )

- FAST_START_PARALLEL_ROLLBACK dynamic
parameter(initial parameter or alter system)
parallel rollback .

- maximum parallel recovery servers
processes .
1. FALSE : .
2. LOW : 2 * CPU_COUNT
3. HIGH : 4 * CPU_COUNT



Monitoring parallel rollback

- V$FAST_START_SERVERS : fast start parallel rollback
query slaves information

- V$FAST_START_TRANSACTIONS : fast start rollback
transactions status information
(Transaction ID, Undo blocks done, Undo blocks total, CPU
time in seconds)

Fast-Start On-demand rollback

- dead transaction lock held SMON
parallel rollback transaction
rollback data block access users
waiting .

- On-demand block recovery user access data
block initiate rollback (immediately recover data
blocks accessed by user transactions) the remainder of the
blocks background recover .

CF. , Fast-Start Rollback Enable .



Block detection and Repair

Oracle Support .

Some facilities for detecting and recovering from block
corruptions

- DB_VERIFY utility : offline file check
both online and backup datafiles
online performance .
(from Oracle7)

- DB_BLOCK_CHECKING : initial parameter setting
blocks check .
(new from Oracle8i)

- ANALYZE command : SQL command analyze table
validate structure syntax invalid rows in partitioned
tables check .
(changed from Oracle8i)

- DBMS_REPAIR : new package PL/SQL interface block
corruptions detect and repair .
package install . (by ‘SYS’)
$ORACLE_HOME/rdbms/admin/dbmsrpr.sql
$ORACLE_HOME/rdbms/admin/prvtrpr.plb
(new from oracle8i)

DBMS_REPAIR
1. connect as user ‘SYS’

2. block corruption check result table
DBMS_REPAIR.ADMIN_TABLE(
‘REPAIR_TABLE’, DBMS_REPAIR.REPAIR_TABLE,
DBMS_REPAIR.CREATE_ACTION,
’TS_REPAIR_DAT’);
>> check table name
>> table data repair_table, index data orphan_table
>> action : table .
(purge_action : delete data from objects,
drop_action : drop the table)
>> table tablespace

3. check corruptions
( 3 parameters )
DBMS_REPAIR.CHECK_OBJECT (‘SCOTT’, ’EMP’,
CORRUPT_COUNT=>:ccnt) ;
>> schema name
>> object name
>> corrupted blocks return



Others

- initial parameter DB_BLOCK_CHECKING=TRUE block
corruption ORA-1578 error trace files
dump .

- Performance reading
check .

- Analyze table valid structure command partitioned
tables invalid rows check

Making Objects Usable

- DBMS_REPAIR.FIX_CORRUPT_BLOCKS(’SCOTT’, ‘EMP’,
FIX_COUNT=>:fcnt)
software corruption block mark
( REPAIR_TABLE )

- DBMS_REPAIR.SKIP_CORRUPT_BLOCKS(’SCOTT’, ‘EMP) ;
corruption marked blocks data read
.
(DBA_TABLES SKIP_CORRUPT )

- Skipping blocks ?
1. inaccessible
2. marked blocks point indexes corrupt
3. referential integrity constraints may be violated
disable and re-enable constraints to identify violations
4. freelist corrupt dbms_repair.rebuild_freelists
freelist initialize .

Checking index entries pointing to rows in corrupt data blocks

1. check table
DBMS_REPAIR.ADMIN_TABLES(
’ORPHAN_TAB’, DBMS_REPAIR.ORPHAN,
DBMS_REPAIR.CREATE_ACTION’, ‘TS_REPAIR_IND’) ;
>> corrupt data table >> index data check
>> table >> table tablespace

2. check index for orphan keys
DBMS_REPAIR.DUMP_ORPHAN_KEYS(‘SCOTT’, ‘EMP’,
ORPHAN_TABLE_NAME=>’ORPHAN_TAB’,
KEY_COUNT=> :kcnt) ;
>> schema name >> object name >> table name
>> number of keys identified

3. table corrupt index rebuild
select index_name from orphan_tab where table_name =
‘EMP’ ;



Limitations of DBMS_REPAIR
- out-of-line columns ignored

- IOT and LOB indexes

- Bitmap and function-based indexes

- Key values 3,950 bytes



14. Features of Net8
Oracle8
New Features .

Net8 and the listener

Net8 and the listener : new parameters

- Oracle8i default listener
listener.ora file . listener start
instance start PMON listener
service handler .
, instance listener start ( ,
listener.ora file instance listener.ora
file )
(every 1 minute) .
CF. listener.ora file
. , listener port location
listener.ora file .

- New initial parameter
INSTANCE_NAME = NIS8I (default SID)
SERVICE_NAMES = NIS8I_A1, NIS8I_B1

- instance name instance name
multiple instances common services share
uniquely instance name identify .

- instance service names listener service
name .
, random listener access service
name .
NIS8I listener
, NIS8I_A 1 listener NIS8I_B
1 listener 2 listener
client .

CF. instance NIS8I connect
(connection request ) client tnsnames.ora
connection information SERVICE_NAME initial
parameter service_names . ,
service_names default (typically global
database name) db_name.db_domain db_domin
parameter default ‘WROLD’ Oracle8.1.5 manual
Oracle8.1.6.2 parameter setting
null . , SID = INSTANCE_NAME = DB_NAME
SERVICE_NAMES default SID .



Load balancing

- connection load balancing
.
1. number of active connections among various instances
2. dispatchers for the same service

- load balancing order
1. client load balancing and failover : random
listener available next available listener
.
2. listener all available nodes information
least CPU usage nodes .
3. listener (MTS ) connection
dispatcher information route .

- Oracle Meta link “Note:75712.1”

1. Client load balancing and failover
tnsnames.ora : client listener
FAILOVER LOAD_BALANCE
listener . load_balance on
listener random off
sequential .
CF. LOAD_BALACE=ON FAILOVER=OFF
listener address random 1
failover .

NIS8I =
(DESCRIPTION =
(FAILOVER=on) (LOAD_BALANCE=on)
(ADDRESS = (PROTOCOL = tcp) (HOST = 203….106) (PORT =
1521))
(ADDRESS = (PROTOCOL = tcp) (HOST = 203….104) (PORT =
1521))
(CONNECT_DATA = (SERVICE_NAME = NIS8I_A))
)
CF. Version “ADDRESS_LIST =” parameter
.

- tnanames.ora listener
listener
1.
2. instance
3. instance down
, listener process
failover error return
.



- OPS SID SERVICE_NAME,
INSTANCE_NAME .
NIS8I =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = tcp) (HOST = 203….106) (PORT
= 1521))
(CONNECT_DATA = (SERVICE_NAME = NIS8I_A)
(INSTANCE_NAME = NIS8I))
)
CF. instance_name parameter parallel server .
, OPS instance_name parameter
.

2. Connection and Node load balancing (MTS only)

- listener .
instance OPS
instance listener
connection load balancing .

- , instance local remote listener
client connection request
. >> connection load balancing

- client random listener service name or sid
connection request listener CPU
usage server(instance) connect
. >> node load balancing

CF. listener listener process
start . , listener.ora file .

CF. service handlers mts_dispatchers
dispatcher .

CF. Test remote listener dispatcher local
listener dispatchers + local server dedicated
.
, local server dispatcher 6 7 service
handlers .

CF. >>
$ lsnrctl services listener_name



3. dispatcher routing

- MTS listener instance
listener listener .

- Multiple node instance SID
SERVICE_NAMES .

- instance MTS
instance listener
connection load balancing initial
parameter MTS_DISPATCHERS listener
information listener connection
listener current status information
setting . listener CPU
usage node connection
dispatcher process connection
.

CF. instance start PMON dispatcher
information listener register (monitor
the system every 30 seconds by default) information
10 update .
( , 1 check 10 node
heavy transaction update )

CF. Oracle8 connection dispatcher
connection Oracle8i active connection
node dispatcher .

>> 3 load balancing connection
client server side process dispatcher process load
balancing .



Scalability

New parameters

- PMON default listener service
initial parameter local_listener default value
.

- LOCAL_LISTENER default local host port 1521,
protocol tcp ipc . port
default value parameter
.

1. * 1522 port ?”
listener.ora :
LISTENER = (ADDRESS_LIST = (ADDRESS= (PROTOCOL=
TCP) (Host= 203.246.130.104) (Port= 1522)))

init$SID.ora :
local_listener =
"(address = (protocol=tcp) (host=203.246.130.104) (port=1522)) "

2. server side tnsnames.ora alias
init$SID.ora
local_listener = lsn_local

tnsnames.ora :
lsn_local =
(address = (protocol = tcp)(host = 203.246.130.104)(port=1522))

CF. init$SID.ora local_listener parameter
multiple local listener .

- MTS_DISPATCHERS
connection load balancing listener
mts_dispatchers .
1. mts_dispatchers=”(protocol=tcp)(dispatchers=2)”
local listener .
2. mts_dispatchers=
”(protocol=tcp)(dispatchers=2)(listener=lsn_remote)”
server side tnsnames.ora lsn_remote
remote listener .

- LARGE_POOL_SIZE= ?
parameter Dedicated Server
process session memory sort UGA
MTS Shared server large pool
.



Multiple instance configuration step on MTS

- instance 'NIS8I' 203.111.111.104 local
listener 203.111.111.106 remote listener
.

Step1. Server side tnsnames.ora listener alias
MLTNIS8I = (ADDRESS = (PROTOCOL = tcp) (HOST =
203.111.111.106) (PORT = 1521))

Step 2. initial parameter mts_dispatchers (remote and local).
mts_dispatchers =
"(PROTOCOL=TCP)(DISPATCHERS=3)(listener=MLTNIS8I)"
mts_dispatchers = "(PROTOCOL=TCP) (DISPATCHERS=3)"

Step 3. optional : SID service name
service_names = NIS8I_A

Step 4. local server listener start
Step 5. remote server (203.111.111.106) listener start
Step 6. database startup
Step 7. client tnsnames.ora setting
NIS8I = (DESCRIPTION = (FAILOVER=on)
(LOAD_BALANCE=on)
(ADDRESS = (PROTOCOL = tcp) (HOST = 203.111.111.104)
(PORT = 1521))
(ADDRESS = (PROTOCOL = tcp) (HOST = 203.111.111.106)
(PORT = 1521))
(CONNECT_DATA = (SERVICE_NAME = NIS8I_A)))

Step 8. “$ lsnrctl services” command listener
.

CF. listener port 1521 port
local listener server side tnsnames.ora
mts_dispatchers .
CF. 203.111.111.106 instance multiple
instance node load balancing
(service_names SID
) 203.111.111.106 instance
.

, listener port . listener port
multiple instance
instance remote listener .

CF. node load balancing listener process
start instance instance
listener start listener
process instance service handler .



Multiple presentation support

- IIOP, HTTP protocol
MTS .
Dedicated protocol .

- MTS provides the ability to configure a dispatcher for each
presentation.

- JAVA support :
1. default TTC presentation
( standard Net8 clients and java stored procedures)
2. Enterprise Java Beans : GIOP presentation
3. CORBA servers : GIOP presentation

- Oracle standard presentation TTC(Two Task
Common) default MTS
dispatcher multiple presentation
.

- GIOP(General Inter-Orb Protocol) IIOP(Internet Inter-Orb
Protocol) EJB and CORBA Server connections
presentation protocol .

- Java stored procedures MTS Dedicated server mode
EJB and CORBA server
CORBA IIOP MTS .

- presentation layer setting TTC
GIOP setting mts_dispatchers parameter .

- Initial parameter init$SID.ora
mts_dispatchers = “(PROTOCOL=TCP)(PORT=1820)
(DISPATCHERS=1)(PRE=ttc)"
mts_dispatchers = "(PROTOCOL=TCP)(PORT=2332)
(DISPATCHERS=1) (PRE=oracle.aurora.server.SGiopServer)"

Connection pooling (MTS only)

- Oracle8 New Features connection
pooling resource idle session
resource new session resource
“physical network connection”
resource utilization feature .

- MTS_DISPATCHERS=”(PRO=tcp) (CON=10) (DIS=3)
(POO=on) (TIC=3)(SESS=30)”

- CON=10 : dispatcher connection
POO=on : connection pooling enable
TIC=3 : idle session * 10 , 30
SESS=30 : idle session session



Connection concentration (MTS only)

- client connect request listener
connection manager listener
dispatchers 1 physical link
multiple logical sessions . (Multiplexing)

CF. connection manager
MTS Dedicated mode .
connection concentration
.

- Connection manager
1. tnsnames.ora sample for client connection
NIS8I = (DESCRIPTION = (SOURCE_ROUTE=yes)
Connection manager address (ADDRESS = (PROTOCOL = TCP)
(HOST = 203.111.111.104) (PORT = 1630))
Listener address (ADDRESS = (PROTOCOL = TCP)
(HOST =203.111.111.106)(PORT = 1521))
(CONNECT_DATA = (SERVICE_NAME=NIS8I_A))
2. cman.ora sample for server configuration
cman =
(ADDRESS_LIST=
(ADDRESS=(PROTOCOL=tcp)(HOST= 203.111.111.104)
(PORT=1630) (QUEUESIZE=32)))
cman_admin =
option (ADDRESS=(PROTOCOL=tcp)(HOST= 203.111.111.104)
. (PORT=1830))
cman_profile = (parameter_list=
(MAXIMUM_RELAYS=1024)
(LOG_LEVEL=1) (TRACING=no)
(RELAY_STATISTICS=yes) (SHOW_TNS_INFO=yes)
(USE_ASYNC_CALL=yes)
(AUTHENTICATION_LEVEL=0)
(REMOTE_ADMIN=FALSE))
3. initial parameter mts_dispatcher sample
mts_dispatchers=”(protocol=tcp) (dis=3) (mul=on)”
CF. multiplex=on setting connection
concentration enable .

- lsnrctl command cmctl command .
$ cmctl start
connection manager connection request
cmadmin process register ‘cmgw’ process
cmgw registration cmgw, listener
information names server oracle
names server address information names
server update ‘cmadmin’ process .



Multi protocol interchange(MPI)

- client/server different type protocol .
TCP/IP client SPX Oracle
server connect TCP Oracle connection
manager connect Oracle connection manager
SPX database server connect client
.

- tnsnames.ora for multi protocol interchange
NIS8I = (DESCRIPTION = (SOURCE_ROUTE=yes)
connect to CMAN using TCP (ADDRESS = (PROTOCOL = TCP)
(HOST = 203.111.111.104) (PORT = 1610))
connect to database using SPX (ADDRESS = (PROTOCOL = SPX)
(SERVICE = MPI_LST))
Service name for database (CONNECT_DATA = (SERVICE_NAME=NIS8I_A))

Security

Security features
- client connection access control

- secure access through the Web

- works with Oracle Advanced Security Option

Connection manager

- client server access control rules
. access control connection manager
CMAN.ora CMAN_RULES .

- CMAN_RULES sample
CMAN_RULES = (RULE_LIST =
(RULE=(SRC=204.111.111.101) (DST=203.111.111.104)
(SRV=NIS8I_B) (ACT=accept))
(RULE=(SRC=204.111.111.102) (DST=203.111.111.106)
(SRV=NIS8I_A) (ACT=reject)) )

- client 204.111.111.101 server 203.111.111.104
NIS8I_B database connection request
client 204.111.111.102 server 203.111.111.106 NIS8I_A
database connection request .

- SRC : source host name or IP (client)
DST : destination host name or IP (server)
SRV : target database service or SID (8i SID only)
ACT : ACCept or REJect



Manageability

Oracle8 New Features Page 93 “Advanced
functions of Net8” homepage
http://sh.hanarotel.co.kr/~ggola Net8i network
configuration

Configuration and trace tools

- Configuration : Net8 Assistant tool GUI
.
$ $ORACLE_HOME/bin/netasst

- Tracing : more readable trace text
$ $ORACLE_HOME/bin/trcasst –o sqlnet.trc

Naming services method

1. local naming : client tnsnames.ora

2. host naming : DNS /etc/hosts alias
host name
(TCP/IP, one host only one Instance )

3. centralized naming : names server

4. connection manager names server

5. external naming : third-party name resolution mechanism

ex) CDS, NOVELL, NIS



15. SQL*Plus, PL/SQL and National Language Support

SQL*Plus

New Features

- server manager sql*plus .

- version server manager
.

SQL*Plus syntax review

- create type, create library new
line ‘/’ .

- SQL blank line .

- “#” comment .
comment “-“ “—“ .

- SQL new line “-“
.

- “&” . “&”
escape character
. Default off “”
( “?” )
SQL> set escape ?
.
EX. select ‘?&k’ from dual ; &k .
“?” &k .

External Procedure (Oracle8 NF)

Orace8i external procedure
JAVA .

External procedure Oracle8 NF



New Event Triggers

New Trigger Level

- Oracle7 DML(row and statement level) trigger
database schema(user and DDL)
level trigger .

- Database level trigger : DBA role
Schema level trigger : create trigger privilege
user schema level trigger : create any trigger privilege

Enhanced trigger

- Oracle8 ‘instead of’ trigger views, object
views Oracle8i
.

- ‘INSTEAD OF’ trigger nested table
nested table .

EX. create trigger emp_nest_trg instead of insert on
nested table emp_nest of emp_object for each row
…….

New Event & Changed view

- STARTUP : database open fire
SHUTDOWN : instance shutdown fire
SERVERERROR : database error occurs
LOGON, LOGOFF : user log on and log off fire
CREATE, ALTER, DROP : object

- DBA_TRIGGERS
TABLE_NAME : database, schema level NULL
TRIGGER_TYPE : before and after event for system events
BASE_OBJECT_TYPE : database or schema for system events

- DBMS_STANDARD package : additional functions to obtain
attributes within a trigger

CF. dba_triggers table_name NULL value
database trigger or schema trigger .



Example
- Trigger for database startup and shutdown history
‘SYS’ user login
Database startup and SQL> create table instance_info (
shutdown history 2 servername varchar2(20), username varchar2(30),
3 work_type varchar2(10), work_time date) ;
table
Table created.

SQL> !more start_trg.sql
create or replace trigger inst_start_trg
after startup on database
declare
v_user varchar2(30);
begin
select username into v_user from user_users ;
insert into instance_info values (dbms_standard.database_name, v_user,
'STARTUP', sysdate) ;
end;
Database startup and /
shutdown trigger
SQL> !more down_trg.sql
create or replace trigger inst_sdown_trg
before shutdown on database
declare
v_user varchar2(30);
begin
select username into v_user from user_users ;
insert into instance_info values (dbms_standard.database_name, v_user,
'SHUTDOWN', sysdate) ;
end;
/

SQL> col trigger_name for a20
SQL> col table_name for a10
SQL> select trigger_name, table_name, base_object_type, trigger_type
2 from dba_triggers
trigger 3 where trigger_name in ('INST_START_TRG','INST_SDOWN_TRG');

TRIGGER_NAME TABLE_NAME BASE_OBJECT_TYPE TRIGGER_TYPE
-------------------- ---------- ---------------- -------------------------------------------------------------
INST_SDOWN_TRG DATABASE BEFORE EVENT
INST_START_TRG DATABASE AFTER EVENT

Database shutdown Database shutdown and startup
and startup
SQL> alter session set nls_date_format = 'YYYYMMDD HH24:MI:SS' ;
Session altered.

SQL> col username for a10
SQL> col servername for a15
SQL> select * from instance_info ;

SERVERNAME USERNAME WORK_TYPE WORK_TIME
--------------- ---------- ---------- -------------------------------------------------------
HJSENT.WORLD SYS SHUTDOWN 20010908 12:31:00
HJSENT.WORLD SYS STARTUP 20010908 12:32:44



Autonomous Transactions

Independence

- autonomous transaction PL/SQL blocks call
transaction independent transaction
.

- , main transaction call sub program
nested transaction main transaction commit
and rollback .

- Oracle8i independent
transaction autonomous transaction
.

CF. Oracle7 DBMS_PIPE
.

- PRAGMA setting
.

- PL/SQL block stored procedure,
package/type method, local procedure/function
. , declare section
.

- transaction lock resource main
transaction resource . dead
lock error return .

- Autonomous transaction application
.

Example

- emp_prag1 emp_test 1 row insert commit
autonomous transaction emp_prag2 1 row insert and
call emp_prag2 and 1 row insert commit
or rollback

- emp_prag2 session 3
rows session emp_prag1
data 1 row .



Program example

Autonomous transaction program description :
create or replace procedure emp_prag1 (a_empno number)
is
v_t_odate date ;
pragma autonomous_transaction;
begin

v_t_odate := sysdate ;

insert into emp_test values (a_empno, 'X1', 'Y1', 888, v_t_odate, 10, 10, 11) ;
commit ;

end;
/

Main transaction program description :
create or replace procedure emp_prag2 (a_empno number)
is
v_t_odate date ;
begin
v_t_odate := sysdate ;
insert into emp_test values (a_empno + 1 , 'X1', 'Y1', 888, v_t_odate, 10, 10, 11) ;

-- call procedure for 1 row insertion
emp_prag1(a_empno) ;

insert into emp_test values (a_empno + 2, 'X1', 'Y1', 888, v_t_odate, 10, 10, 11) ;
end;
/

- Session 1
SQL> exec emp_prag2(999) ;
SQL> select * from emp_test ;
1000, 999, 1001 empno data 3
.

- Session2
SQL> select * from emp_test ;
999 empno data 1 .

- Session1
SQL> rollback ;
session1 session2 empno 999 data
1 .

CF. SQL*Plus autocommit on setting
transaction commit
autocommit off .



PL/SQL

Enhancements

- very large package support : maximum package size
.

- Bulk binds : FORALL collection elements
SQL bulk binding .
collection elements(index-by table, nested table, varray, host
array) first last .
……………………
FORALL i in 1..500 LOOP
delete from emp where empno = empnos(i) ;
……………………..
CF. This functionality reduces communication overhead.

- Dynamic SQL : DBMS_SQL package
dynamic SQL
PL/SQL directly .

- Dynamic SQL DML DDL .
Dynamic SQL sample
……..
sql_stm := ‘insert into dept values (:1, :2, :3)’ ;
execute immediate sql_stm using v_dept, v_name, v_loc ;
……
v_where := ‘ where deptnm = ‘ || v_deptnm ;
execute immediate ‘delete from ‘ || v_tab_name || v_where ;
…….
execute immediate ‘create table emp (empno number(4))’ ;

- Oracle8i object features REF-based operations
PL/SQL “UTL_REF” package support

- UTL_REF package contains procedures to support reference-
based operations (Oracle8i manual “Supplied Packages
Reference, Volumne2 “ UTL_REF )

- monitoring and analysis of Program Execution
packages new three APIs tracing, debugging,
profiling .
CF. Oracle8i New Features
Example .
.



Parameter passing by reference

- PL/SQL in or out or in out ,
value parameter reference
.

- NOCOPY mode . , out
parameter procedure reference
parameter pointer .

- sample .

SQL> declare
n number := 100 ;
procedure do_print( n1 in number, n2 in out number, n3 in out nocopy
number) is
begin
n2 := 200 ;
dbms_output.put_line(n1) ; -- print 100
n3 := 300 ;
dbms_output.put_line(n1) ; -- print 300
end ;
begin
do_print(n, n, n) ;
dbms_output.put_line(n) ; -- print 200
end ;
/
100
300
200
SQL>

- 100 ( input)
300 (nocopy mode) 200 ( out
variable) print .

CF. 300 nocopy mode 200 in out
.



National Language Support

Oracle8 New Features
.

character set

- Database Creation , create database
character set national character set
.

SQL> Create Database DBNAME Logfile Group 1… Datafile…
Character Set US7ASCII
National Character Set KO1601KSC ;

- National character set data
NCHAR, NCLOB, NVARCHAR2 National
.

Dual currency support (for Euro)

- 1999 1 EU new currency Euro
.

- euro sample .
SQL> alter session set nls_dual_currency=eur;
Session altered.

SQL> select to_char(888, 'U999') from dual ;

TO_CHAR(888,'U
----------------------------
EUR888

NLS sort

- sort function based index
.

- SQL> create index xak_nls_ename on emp (nlssort(ename,
‘NLS_SORT = German’)) ;

- SQL> select * from emp where ename > ‘Ulrike’ ;

- German ‘z’ < ‘ä’ . NLS_COMP (dynamic
initialization parameter) parameter linguistic comparisons
binary default language for the session German
setting .



16. Database Security

Password Management


Scripts $ORACLE_HOME/rdbms/utlpwdmg.sql

Invoker’s Rights

PL/SQL execution user

- Oracle8i version PL/SQL block user
PL/SQL user .

- Oracle8i AUTHID DEFINER
CURRENT_USER .

- DEFINER default PL/SQL block
user CURRENT_USER
block user .

- sample akman user procedure scott
user .

Sample procedure and test result

- User ‘akman’ procedure

create or replace procedure inv_t1 (av_tname in varchar2)
authid current_user is
v_stmt varchar2(40) default 'drop table ' || av_tname ;
begin
execute immediate v_stmt ;
end;
/

create or replace procedure inv_t2 (av_tname in varchar2)
-- definer default .
authid definner is
is
v_stmt varchar2(40) default 'drop table ' || av_tname ;
begin
execute immediate v_stmt ;
end;
/



SQL> sho user
USER is “AKMAN”
SQL> grant execute on inv_t1 to scott;
SQL> grant execute on inv_t2 to scott;

SQL> create table emp_test as select * from emp;
Table created.
SQL> exec akman.inv_t2('emp_test') ;
BEGIN akman.inv_t2('emp_test') ; END;
*
ERROR at line 1:
User ‘AKMAN’ emp_test
ORA-00942: table or view does not exist
table . ORA-06512: at "AKMAN_T2", line 6
procedure akman error returns

SQL> exec akmanager.inv_t1('emp_test') ;
Procedure inv_t1 PL/SQL procedure successfully completed.
‘AKMAN’ SQL> desc emp_test
user ‘SCOTT’ ERROR:
drop table ORA-04043: object emp_test does not exist
procedure scott

User tracing

- AUTHID PL/SQL block user
.
CF. AUTHID CURRENT_USER block
user_users query username
.

Security policies

Application security

- application privileges and
roles security policy
privileges and roles application
.

Application context
- Application context : set up session-based attributes securely.
session level , access user information retrieve
and store security policy.

Fine-Grained access control
- Find-Grained access control : implement security policies as a
low level of granularity.



Application Context

Features
Application-specific - Each application can have its own unique context, with
different attributes.
order entry application access control
customer number, position, and sales region attributes
user
, , .
Secure
- Application context can provide security in multiple ways.
order entry user customer number
valid
user application context
.

- Context context implementing
“trusted package” .
Flexible
- within a function implementing a security policy to determine
the correct predicate to return
select * from emp application
context position different predicates(WHERE
clause) return . , manager
data data
.

- Context attribute can be used within the predicate itself.
, Predicate(as a bind variable) return .
predicate ‘where cust_num =
sys_context(‘order_entry’, ‘cust_num’)’ return .

Application Context creation and test

1. application rule .
emp table connect user emp ename
row access .

2. application context function
PL/SQL package

3. application context

4. package context test

CF. Oracle8i
. Manual .



Example

- context user create any context
privilege . (drop drop )
SQL> conn system/manager
SQL> grant create any context to scott ;
Connected.
SQL> select * from emp ;
EMPNO ENAME JOB MGR HIREDAT SAL COMM DEPTNO
-------------------------------------------------------------------------------------------------------------------------
7369 SCOTT CLERK 7902 17Dec80 800 40
7499 ALLEN SALESMAN 7698 20Feb81 1600 300 30
7521 WARD SALESMAN 7698 22Feb81 1250 500 30
8970 MILLER CLERK 7782 23Jan82 1300 10
9889 CUSTOMER GENERAL 7369 21Jun01 600 200 40

- context package
SQL>
create or replace package apl_secure_context
is
procedure set_secure_emp ;
end ;
/
Package created.
SQL>
create or replace package body apl_secure_context
is
procedure set_secure_emp
is
v_empno emp.empno%type ;
begin
select empno into v_empno from emp
where ename = sys_context('USERENV', 'SESSION_USER') ;
DBMS_SESSION.SET_CONTEXT('app_emp_context','empno',v_empno);
end;
end;
/
Package body created.

- context creation and test
SQL> create or replace context app_emp_context using apl_secure_context ;
Context created.
SQL> select sys_context('USERENV', 'SESSION_USER') USERNAME from dual ;
USERNAME
---------
SCOTT

SQL> exec apl_secure_context.set_secure_emp ;

SQL> select sys_context('APP_EMP_CONTEXT', 'EMPNO') EMP_NUMBER from dual ;
EMP_NUMBER
----------
7369



Fine-Grained Access control

Fine-grained access control provides a way to extend table-
based or view-based security to finer levels of granularity than
previously possible.

Merits

- application security
data server security
costs .
CF. application user data
access server fine-grained access
control . (Row level security )

- application level security
, user security
security .

Features

- table-based or view-based security policies : table or view
flexible implementation
salary emp columns
emp_view application fine-
grained access control
application . select, insert, update,
and delete .

- Multiple policies per table
table policies
logically ‘AND’ .

- Highly scalability
application context attributes
values parsed bind variable .
parsed optimization
.

How it works

- direct or indirect (subquery ) access
automatically invokes the policy

- policy package predicate(a where condition) return
Database dynamically rewrites the SQL statement by
appending the predicate

- dynamically rewrite bind variable
re-parsing or re-optimizing .



Implementing

- fine-grained access control secure policy
application context
.

- fine-grained access control application context
application context policy
packages .

- application context fine-
grained access control .

CF. DBMS_RLS Package
$ORACLE_HOME/rdbms/admin/dbmsrlsa.sql script
.dbms_rls package
script .
(RLS : Row Level Security Administrative Interface)

Example

- login user sales table select
row Row Level Security Policy

1. building application context (optional)
application context .
(1) context function package
page application context example
(2) context
page application context example
(3) policy data
SQL> grant select on emp to customer ;
SQL> grant select on sales to customer ;
SQL> select * from emp ;

EMPNO ENAME JOB MGR HIREDAT SAL COMM DEPTNO
---------------------------------------------------------------------------------------------------------
9889 CUSTOMER GENERAL 7369 21Jun01 600 200 40
7369 SCOTT CLERK 7902 17Dec80 800 40
7499 ALLEN SALESMAN 7698 20Feb81 1600 300 30
7521 WARD SALESMAN 7698 22Feb81 1250 500 30
8970 MILLER CLERK 7782 23Jan82 1300 10

SQL> select * from sales;

EMPNO DEPT ITEM STORE SALES_DATE
--------------------------------------------------------------------------------
9889 40 435 Prince 21Jun01
7369 40 839 Head 17Dec80
7499 30 630 Wilson 20Feb81
7521 30 463 Dunlop 22Feb81
8970 10 693 Unsquahable 23Jan82



2. building policy packages
create or replace package secure_emp_fgrain
is
function secure_emp_fgrain_fnc(d1 varchar2, d2 varchar2) return varchar2 ;
end ;
/

create or replace package body secure_emp_fgrain
is
function secure_emp_fgrain_fnc(d1 varchar2, d2 varchar2) return varchar2
is
v_predicate varchar2(2000) ;
begin
apl_secure_context.set_secure_emp ;
v_predicate := 'empno = sys_context('APP_EMP_CONTEXT', 'EMPNO')' ;
return v_predicate ;
end;
end;
/

3. associate policy packages with tables and views

SQL> exec dbms_rls.add_policy(‘scott’,’sales’,’sales_policy’,’scott’,
’secure_emp_fgrain.secure_emp_fgrain_fnc', 'select')

CF. line .

4. Test
SQL> select * from sales;

SCOTT SALES EMPNO DEPT ITEM STORE SALES_DATE
. ---------------------------------------------------------------------
7369 40 839 Head 17Dec80

SQL> select * from sales where dept=30;
DEPT = 30 POLICY no rows selected
PREDICATE AND
. SQL> conn customer/customer11
SQL> select * from scott.sales ;

CUSTOMER EMPNO DEPT ITEM STORE SALES_DATE
SALES . ------------------------------------------------------------------------
9889 40 435 Prince 21Jun01



DBMS_RLS package

- Policy :
ADD_POLICY (object_schema, object_name, policy_name,
function_schema, function_name, statement_type,
[update_check], [enable])
CF. update_check default = ‘FALSE’ (insert or update
statement argument ‘TRUE’ insert or
update policy
check )
enable default = ‘TRUE’

- Policy :
DROP_POLICY(object_schema, object_name, policy_name)

- Policy enable, disable :
ENABLE_POLICY(object_schema, object_name, policy_name,
[enable])
CF. enable default = ‘TRUE’

- Policy Re-parsing :
REFRESH_POLICY(object_schema, object_name, policy_name)



17. Partial addition for Oracle8i

Constraints

Oracle7 Oracle8 constraints
features .

Deferred constraint checking

- constraint
constraint not deferrable (default)
deferrable initially immediate (default)
initially deferred

Oracle7 constraint : constraint check

Oracle8 New features : SQL check using set
constraint statement

Oracle8i New features : commit time check

Constraint
- set constraint command : apply only to current transaction
SQL> set constraint(s) all/constraint_name
deferred/immediate ;

- alter session set constraint command : apply to current session
SQL> alter session set constraint(s) = deferred/immediate ;

Example
- default constraint(not deferrable) test
SQL> select * from dept ;
DEPTNO DNAME LOC
---------- -------------- -------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
40 OPERATIONS BOSTON
SQL> alter table emp add constraint xfk_emp_dept foreign key (deptno) references dept(deptno) ;
SQL> insert into emp (empno, deptno) values (7777, 50) ;
ERROR at line 1:
ORA-02291: integrity constraint (SCOTT.XFK_EMP_DEPT) violated - parent key not found

SQL> set constraint XFK_EMP_DEPT deferred ;
ERROR at line 1:
ORA-02447: cannot defer a constraint that is not deferrable



- deferrable initially immediate test
SQL> alter table emp drop constraint xfk_emp_dept ;
SQL> alter table emp add constraint xfk_emp_dept foreign key (deptno) references dept(deptno)
deferrable initially immediate ;
ERROR at line 1:

SQL> set constraint XFK_EMP_DEPT deferred ;
1 row created.
SQL> commit ;
ERROR at line 1:
ORA-02091: transaction rolled back

- deferrable initially deferred test
SQL> alter table emp drop constraint xfk_emp_dept ;
SQL> alter table emp add constraint xfk_emp_dept foreign key (deptno) references dept(deptno)
deferrable initially deferred ;
1 row created.
SQL> commit ; set constraint command commit check .
ERROR at line 1:
ORA-02091: transaction rolled back
SQL> set constraint XFK_EMP_DEPT immediate ;
ERROR at line 1:

constraint enable/disable & validate/novalidate

- constraint enable/disable option data
check .

- Validate : data check
Novalidate : data no-check
CF. option default ‘validate’ .

- feature nonunique index Primary or
Unique constraint automatic index
.

- , PK or UK constraint disable
automatic index drop enable automatic index
rebuild .



- FK data
.

SQL> select * from dept ;
DEPTNO DNAME LOC
---------- -------------- --------------------------
20 RESEARCH DALLAS
30 SALES CHICAGO
SQL> alter table emp drop constraint XFK_EMP_DEPT ;
1 row created.
SQL> alter table emp add constraint xfk_emp_dept foreign key (deptno)
2 references dept(deptno) enable novalidate ;
Table altered.
*
ERROR at line 1:

SQL> alter table emp modify constraint xfk_emp_dept enable validate ;
*
ERROR at line 1:
ORA-02298: cannot validate (SCOTT.XFK_EMP_DEPT) - parent keys not found
SQL> alter table emp modify constraint xfk_emp_dept disable validate ;
*
ERROR at line 1:
SQL> alter table emp modify constraint xfk_emp_dept disable novalidate ;
Table altered.

1 row created.

*
ERROR at line 1:

SQL> alter table emp modify constraint xfk_emp_dept enable novalidate ;
Table altered.

SQL> delete from emp where deptno = 50 ;
2 rows deleted.

SQL> alter table emp modify constraint xfk_emp_dept disable validate ;
Table altered.

SQL> insert into emp (empno, deptno) values (7777, 50) ; DML .
*
ERROR at line 1:
ORA-25128: No insert/update/delete on table with constraint
(SCOTT.XFK_EMP_DEPT) disabled and validated
Table altered.



- index non-unique index
primary key constraint automatic
index index .

- Oracle7 index column
PK or UK .
SQL> alter table department add constraint xpkdepartment primary key (deptno) ;
SQL> select constraint_name, status from user_constraints where table_name = 'DEPARTMENT';
CONSTRAINT_NAME STATUS
------------------------------ --------
XPKDEPARTMENT ENABLED
SQL> select index_name, status from user_indexes where table_name = 'DEPARTMENT' ;
INDEX_NAME STATUS
------------------------------ -------- disable constraint command
XPKDEPARTMENT VALID
SQL> alter table department modify constraint XPKDEPARTMENT disable ;
------------------------------ --------
XPKDEPARTMENT DISABLED
no rows selected
SQL> alter table department drop constraint xpkdepartment ;
no rows selected
SQL> create index xnk_dept_loc on department (deptno, loc) ;
no rows selected
INDEX_NAME STATUS
------------------------------ --------
XNK_DEPT_LOC VALID
------------------------------ --------
INDEX_NAME STATUS
------------------------------ --------
XNK_DEPT_LOC VALID
SQL> select * from department ;
DEPTNO DNAME LOC
----------- -------------- - ------------
20 RESEARCH DALLAS
30 SALES CHICAGO

SQL> insert into department values ('40', 'DBA', 'Cyber') ;
*
ERROR at line 1:
ORA-00001: unique constraint (SCOTT.XPKDEPARTMENT) violated



SQL> alter table department drop constraint XPKDEPARTMENT ;
SQL> insert into department values ('40', 'DBA', 'Cyber') ;
1 row created.
*
ERROR at line 1:
ORA-02437: cannot validate (SCOTT.XPKDEPARTMENT) - primary key violated
SQL> delete from department where deptno=40 and loc='Cyber' ;
1 row deleted.
Table altered.
SQL> alter table department disable constraint xpkdepartment ; modify constraint disable command .
Table altered.
------------------------------ --------
XPKDEPARTMENT DISABLED
INDEX_NAME STATUS
------------------------------ --------
XNK_DEPT_LOC VALID
SQL> alter table department enable constraint xpkdepartment ;
Table altered.
------------------------------ --------
INDEX_NAME STATUS
------------------------------ --------
XNK_DEPT_LOC VALID

CF. nonunique index unique index PK
. , unique index PK
PK disable unique index drop enable PK
PK Index . , nonunique index
feature .



Rely/NoRely Flag

- DBA rely flag DBMS constraint
validity check constraint valid
.

- , usually optimizer Materialized View query
rewrite validate constraint
query rewrite
validate rely flag
constraint query rewrite .

- Restrictions
1. RELY and NORELY are relevant only if you are modifying
an existing constraint(that is, you have issued alter table …
modify constraint statement.)
==> manual
. Rely option constraint
“relevant only if” .
2. Not null constraint .

- rely constraint user_constraints
.
SQL> alter table emp modify constraint XFK_EMP_DEPT rely disable -
novalidate ;

Table altered.

SQL> select CONSTRAINT_NAME, STATUS, VALIDATED, RELY
2 from user_constraints where constraint_name = ' XFK_EMP_DEPT ' ;

CONSTRAINT_NAME STATUS VALIDATED RELY
------------------------------ -------- ------------------- -------
XFK_EMP_DEPT DISABLED NOT VALIDATED RELY

SQL> alter table emp modify constraint XFK_EMP_DEPT norely enable;
Table altered.

SQL> select CONSTRAINT_NAME, STATUS, VALIDATED, RELY
from user_constraints where constraint_name = ' XFK_EMP_DEPT ' ;

CONSTRAINT_NAME STATUS VALIDATED RELY
------------------------------ -------- ------------------- -------
XFK_EMP_DEPT ENABLED VALIDATED



Oracle8i object new features


MEMBER Method

- method PL/SQL, 3GL(extended pro*c), Package
body JAVA .

Dangling REFs

- Oracle8 dangling REF
update .
EX) update dept set mgr=NULL where mgr is DANGLING ;

- Oracle8i update command
.
EX) analyze table dept validate ref update set dangling to
null ;

- dangling REFs does not
return an error.

Nested Table Enhancements
- Nested tables may be stored as Index-Organized Tables

- Constraints may be defined on nested tables

- Locators and references can be used to SELECT or perform
DML on nested tables.(REF ) A locator returns a
pointer to the nested table, not the rows of the table.

Table function for Nested table and VARRAY

- Table function (collection type table )
Insert into ord values (100, line_item_array_type(
line_item_type(300, 2300),
line_item_type(700, 3300) )) ;

Select o.ord_id, i.prod_id, i.price
Table function
nested table elements
From ord o, TABLE(o.line_items) I
DML . Where i.prod_id in (300, 700) ;
Varray PL/SQL procedure
. Update
TABLE( select o.line_items from ord o where ord_id = 100) i
Set i.quantity = 100
Where prod_id = 400 ;



LOBs


CF. internal LOB storage CACHE/NOCACHE,
LOGGING/NOLOGGING .
[CACHE NOLOGGING ]

DBMS_LOB.CREATETEMPORARY (except BFILE type)

- DBMS_LOB.CREATETEMPORARY command
user’s temporary tablespace temporary lob
.(except BFILE)

- database LOB data copy application
ability design .( sample )

- Temporary LOB the session, transaction, or call
no rollback and no redo logging generated

- In SQL, temporary LOBs may only be used in WHERE clauses,
an INSERTed value, or as the assigned value in an UPDATE.

- temporary lob lob data
insert .

-- The following example uses a temporary LOB to reverse one of the LOB alues in the table
-- and then inserts the reversed LOB as a new row.
declare
c_lob CLOB; -- permanent LOB locator
t_lob CLOB; -- temporary LOB locator
buf varchar2(32000); -- }this example assumes the LOB is
buf2 varchar2(32000); -- }less than 32K.
chunk number;
len number;
offset number;
amount number;

begin

select c2 into c_lob from test_lobs where c1 = 1;
-- Create a temporary LOB.
-- The parameters to CREATETEMPORARY are locator, use caching or not and duration.
-- Set no caching and a duration of call
-- since the temporary LOB is not required outside of this PL/SQL block.

DBMS_LOB.CREATETEMPORARY(t_lob, FALSE, DBMS_LOB.CALL); -- NEW 8i

-- Chunk for storing the LOB value.
-- Using this amount for reads and writes of the LOB will improve performance.
chunk := DBMS_LOB.GETCHUNKSIZE(c_lob);
dbms_output.put_line('Chunksize of column c2 is '||chunk);
dbms_output.put_line('Chunksize of temporary LOB is '|| DBMS_LOB.GETCHUNKSIZE(t_lob));



len := DBMS_LOB.GETLENGTH(c_lob);
offset := 1;
buf := null;

while offset < len loop
if len - (offset-1) > chunk then
amount := chunk;
else
amount := len - (offset-1);
end if;
buf2 := null;
DBMS_LOB.READ(c_lob,amount,offset,buf2);
buf := buf||buf2;
offset := offset + amount;
end loop;

--Reverse the read data and write it to the temporary LOB.
buf2 := null;
for i in reverse 1..len loop
buf2 := buf2||substr(buf,i,1);
end loop;

-- Write the whole lot in one go.
-- Note, if this was a large amount of data then ideally it should be written using the
-- available chunksize of the temporary LOB.
DBMS_LOB.WRITEAPPEND(t_lob,len,buf2); -- NEW 8i

-- Now insert a new row into the table setting the CLOB column to the value of the temporary LOB.
-- This can be done in one of two ways:
--(i) A new row can be inserted with an empty locator, the locator
-- retrieved and the LOB value copied with DBMS_LOB.COPY.
--(ii) A new row can be inserted passing the temporary LOB locator
-- as a bind variable to the insert.
--
--Using the second method:
insert into test_lobs values (5,t_lob,null,null) returning c2 into c_lob;

--Free the temporary LOB explicitly.
if DBMS_LOB.ISTEMPORARY(t_lob) = 1 then
DBMS_LOB.FREETEMPORARY(t_lob);
end if;

dbms_output.put_line('Length of CLOB inserted into record 5 is '|| DBMS_LOB.GETLENGTH(c_lob));
commit;
end;
/



Recovery Manager


Catalog Database
- Oracle8
# sqlplus rmanuser/rmanuser
SQL> @$ORACLE_HOME/rdbms/admin/catrman.sql
: catalog user .
user default tablespace catalog DB
.

- Oracle8i
RMAN> create catalog tablespace “TS_RMAN” ;
: command catalog DB .

Newly supported command

- RMAN Database startup and shutdown command
.



Advanced Queuing

Oracle8i new features
interface C++ JAVA .

Database Security
- AQ_ADMINSTRATOR_ROLE allows maintain queues in any
schema.

- AQ_USER_ROLE user’s schema queues enqueues and
dequeues .

- DBMS_AQADM allows the user to maintain AQ objects in
their schema only.

- DBMS_AQ allows enqueues and dequeues from queues in
user’s schema only.

- Grant(or revoke)_system_privilege in package
DBMS_AQADM grant or revoke the
ENQUEUE_ANY, DEQUEUE_ANY, MANAGE_ANY
privileges .

- Grant(or revoke)_queue_privilege in package
DBMS_AQADM grant or revoke the
ENQUEUE_ANY, DEQUEUE_ANY, ALL privileges on
individual queues .

Oracle Enterprise Manager V2

Oracle Enterprise Edition OEM
free-of-charge DBA management pack .

New architecture

- 2-tier , fat client Oracle
Management Server 3-tier distributed
architecture .

- Java console administration application
platform .
client JAVA console – middle tier OMS – server Oracle through
agent

- OMS a set of database tables repository
Multiple system .


Oracle History #5

More Related Content

What's hot (20)

Viewers also liked (7)

Similar to Oracle History #5 (20)

More from Kyung Sang Jang (14)

Recently uploaded (20)

Oracle History #5