Database Performance Management in Cloud

International Journal of Engineering and Management Research e-ISSN: 2250-0758 | p-ISSN: 2394-6962
Volume-10, Issue-5 (October 2020)
www.ijemr.net https://doi.org/10.31033/ijemr.10.5.11
46 This work is licensed under Creative Commons Attribution 4.0 International License.
Database Performance Management in Cloud
Navendu Mishra1
, Jayant Sawarkar2
and Suraj Dubey3
1
Student, Department of Computer Science & Engineering, Birla Institute of Technology & Science, Pilani, INDIA
2
Assistant Professor, Department of Computer Science & Engineering, Datta Meghe College of Engineering, Navi Mumbai,
INDIA
3
COO, Osource Global Private Limited, Mumbai, INDIA
2
Corresponding Author: jdsawarkar@yahoo.com
ABSTRACT
Managing large chain of Hotels and ERP database
comprises of core areas such as HRMS & PIP.HRMS
(Human Resource Management System), which further
includes areas such as Soft Joining, Promotion, Transfer,
Confirmation, Leave Attendance and Exit, etc. PIP (Payroll
Information Portal), wherein employees can view their
individual Salary details, submit investment declaration,
Reimbursement claim & CTC structuring, etc. Management
of Large Chain of Hotels and ERP Database in AWS Cloud
involves continuous monitoring with regards to the areas
such as Performance of resource usages and optimization
techniques relating to the use of PL/SQL. High Availability
(HA) of data is accomplished through the Backup and
Recovery mechanism and security of the data by Encryption
& Decryption mechanism.
Broad Academic Area of Work: CLOUD COMPUTING
Keywords-- AWS Cloud, Database, ERP, Hotels, PL/SQL,
S3 Bucket and Query
I. INTRODUCTION
Oracle Database works by utilizing the various
components of the Hardware resources such as CPU,
Memory (RAM), HDD (SAN-STORAGE), Networks, etc.
The performance of the database depends upon the
resource usages. Queries in a database is written to fetch or
get the result, wherein all the resources are utilized.
If the query is a complex query then it will start
consuming the CPU, the Memory and the HDD
(Read/Write - I/O).In such scenarios Query Optimization
or Query Tuning plays a vital role for enhancing the DB
performance.
Most of the organizations prefer to use Oracle
database on the Linux flavors world-wide, however,
Oracle also provides database services for the Windows
Operating System users as well.
Linux is more robust, and from the security point
of view is more desirable as compared to the Windows
O/S.
Oracle database and RAID works very closely.
RAID is known as Redundant Arrays of Inexpensive
Disks. This is used for Load Balancing and data protection
across multiple disks. Most commonly used RAID levels,
are RAID 0, RAID 1, RAID 0+1, RAID 1+0 or 10 and
RAID 5.
RAID 1+0 (RAID 0 and RAID 1), is the best
RAID option or Oracle. For Oracle database control Files,
Online Redo Logs, Archived Redo Logs and Undo
Datafiles, RAID 1+0 is preferred, as need to have multiple
copies of the same. The Temporary datafiles, RAID 0 is
preferred, as if the datafile is lost, the same can be
recreated, as it has no data to restore.
Oracle database Datafiles preferred RAID levels
are RAID 1+0, RAID 0+1 or RAID 5.RAID levels varies
depending upon the price /commercials and performance.
The latest version of 19c provides the database in the form
of CDB & PDB's on ASM.
Oracle's ASM (Automatic Storage
Management) is the most widely used and preferred
solution to provide data striping and mirroring. Here ASM
disks are protected by the hardware RAID, and are defined
using external redundancy and has no failure groups. In
general two disk groups namely DATA and FRA are used.
DATA stores all the database related files and FRA stores
the fast recovery area, which includes multiplexed copies
of online redo logs and control files.
One of the best feature called LVM, which stands
for Logical Volume Manager, is a highly efficient used
framework in the Linux distributions, owing to the facility
of providing logical volume management for the Linux
kernel. The basic function of LVM, is to provide allocation
of disks, mirroring and stripping, and re-sizing the logical
volumes.
Tuning can be done at the Memory level, and also
at the SQL-PL/SQL level. Increasing the CPU will
definitely increase the database server performance.
Also, keeping the datafiles on the SAN box is
mandatory not only for performance (I/O), but also in case
any damage happens to any of the disks due to block
corruption, the other mirrored disks can be used , and
hence, there will not be any data loss or any DB outage.

RAID levels are used for the data redundancy,
owing to the fact that the data is critical.
Oracle GoldenGate is used for the data migration
services to replicate data from one database to the other
database across different geographic regions world-wide.
Oracle GoldenGate allows to migrate the
committed transactions in the database across multiple
heterogeneous systems. Oracle GoldenGate comes as a
separate software product, which needs installation &
configurations.
Database Administrators core responsibility is to
manage all the company’s database pertaining to the very
important and critical areas such as database performance
which utilizes resources, database backup and recovery
(HA), and security through the encryption and decryption
mechanisms, as per the industry standards. The solutions
implemented in the database focuses in the three core areas
of performance, keeping in view the wide variety of clients
world-wide. Before the implementation of any solutions on
the production, On the fore-front of the implementation of
solutions in the OCI (Oracle Cloud Infrastructure) in
conjunction with the AWS cloud. For the migration of
databases to a higher version which presents various
enhanced database features. Another solution which needs
to be implemented for the database is the migration of the
current database from Non-Container database to
container database (CDB / PDB’s), and also to utilize the
ATP (Autonomous Transaction Processing) of Oracle
database. OCI (Oracle Cloud Infrastructure) provides
the Real Application Clustering (RAC) environments,
wherein the automatic storage management is done by
using ASM disks, wherein the High Availability (HA) is
provided continuously, without any database outage
(downtime).This is a very important feature of Oracle
which helps the business to the maximum possible extent.
All the necessary steps are being taken for making sure
that the core areas of the database remain within the
threshold limit, so that the resource usage is not hampered,
and which will automatically enable a better performance
of the database in totality.
AWS RDS/EC2, provides solutions which have been
successfully implemented curtailing to the company’s
business areas world-wide, and which has helped the
company to establish itself as a fore-front in the
Hospitality, ERP, PAYROLL SERVICES, DMS, BE-
SPOKE industry.
II. LITERATURE REVIEW
Database Dictionary and Performance Views
In accordance with the database features, data
dictionary plays a very important role in the Oracle
database, as it stores the meta-data of the database. Meta-
data pertains to the logical structure of tables, views,
indexes, procedure, function, packages, package body,
trigger, etc. A data dictionary can be referred to as
centralized repository of information of data of the
database, which gets updated automatically and which
reflects the changes in the data and structure of the
database, auditing of the DB, grants and privileges.
Figure 1: Oracle database structure
Figure 2: Data dictionary structure
There are basically two types of views which
works when the DB is in OPEN mode. One is V$views,
which are the dynamic performance tables, which are
updated continuously, and presents to the users the data as
of now. A DBA view describes all views in the database.
Parameters in Oracle, plays a very pivotal role in the
overall working of the database.

There are very important views which stores the
Oracle database dictionary information. V$DATABASE
view, shows the overall detailed information about the
database. V$SESSION view, shows the overall sessions
details in the database. The other data dictionary view is
DBA_views, which shows the overall details about various
objects like tables, indexes, SGA (System Global Area),
the physical storage details (data).DBA_DATA_FILES
view stored the database files along with the tablespace
name and their physical location. DBA_SEGMENTS,
specifies the total database size which is actual in use. This
is very important view, as it presents the exact database
size at any point of time. Through this the size of any
schema and any tablespace can be known.
This paper covers the aspects of database
performance, in AWS cloud & Oracle Cloud Infrastructure
(OCI), which is an important factor in the database
management area. The work which has been done to
prepare this paper has helped to a large extent to enhanced
database knowledge not only to the previous versions of
database but also to the latest version of Oracle DB 19c,
which has many newly added database features
(CDB/PDB's)which will definitely help (DBA) and also
the developers.
The implementations of all the features such as
queries to monitor the database CPU consumption
sessions, finding the sessions responsible for the Memory
& I/O consumption has helped a lot to monitor the DB
server performance with great ease, and which has resulted
in the database server performance and health not getting
hampered under any circumstances. Details of each and
every components of database aspects pertaining to the
overall performance of the DB server, as performance is
the most vital part in the database functioning.
III. IMPLEMENTATIONS
CPU, MEMORY & I/O usage monitoring plays
a very important role pertaining to constantly monitoring
the database performance which should always be within
the defined threshold limit, so that the performance of the
database server is not compromised or hampered under
any circumstances. There are events which are related to
the CPU, MEMORY&I/O such as, cursor: pin S, db file
scattered read, db file sequential read, enq: TX - row
lock contention, latch: row cache objects, buffer busy
waits, enq: RO - fast object reuse, direct path read,
library cache: mutex X, etc. These database events are
optimized through the use of various PLSQL techniques,
and which forms the part of the database performance
tuning. Every newer versions of the Oracle database comes
with enhanced database performance techniques, and by
applying which the DB server performance is increased
substantially, which directly increases the application
performance.
The database events which are responsible for the
CPU consumption, hampers the database server
performance which leads to slowness. There needs to be
continuous monitoring of CPU events through the use of
tools like Oracle SQL Developer and TOAD, which have
got much in-built functionality to monitor the health of the
database server. resmgr: cpu quantum is the main event
which needs to be resolved by the process of query tuning
to have the CPU usage within the threshold limit and
thereby enforcing the mechanism that the DB server
performance is not at all hampered. Also up-scaling the
CPU resources during peak hours, also reduces the CPU
contention on the database server.
Memory tuning in Oracle database is the key
factor for the performance of the database server to the
larger extent. MEMORY_MAX_TARGET and
MEMORY_TARGET, are the two MemoryTuning
parameters in Oracle 12c and higher versions of Oracle.
MEMORY_MAX_TARGET is not a dynamic
parameter whereas MEMORY_TARGET is. It means that
out of the available total physical memory, the
MEMORY_MAX_TARGET is the, maximum database
memory which has been assigned from the physical server
memory, and which requires reboot (database bounce) to
take effect of the same in the database, whereas
MEMORY_TARGET is the database memory which can
be assigned to a value equal to that of the
MEMORY_MAX_TARGET
If MEMORY_TARGET, is defined or set, then
Oracle will perform AMM (automatic memory
management) automatically, and both SGA (System
Global Are) & PGA (Program Global Area) are managed
within the allocated memory. There is no need to set
SGA_TARGET &s SGA_MAX_SIZE parameters.
I/O monitoring is also very important as far as the
database server performance is concerned and managing
the I/O, which refers to the Physical reads from and
Physical writes to the SAN storage disks. Also many times
there are scenarios, wherein the same query is being used
again and again, and then in that case the Oracle will read
the data from Memory (Buffer), and which is known as
Soft Parsing, rather than reading from the disks (Hard
Parsing) which directly increase the query performance.
Oracle has inbuilt mathematical functions and
other query clauses which provides the output as desired.
When there is increase in the block changes, then
the Oracle database server performance is hampered.
The increase in the block changes means there is
too much of physical writes directly due to INSERT SQL
statements, during this point DBA has to take the
appropriate action and see that the block changes in the

database is within the threshold limit depending, so that
the database server health is not compromised.
The crux is to get the actual physical block
changes which happens in the database and which results
into the utilization of the available read/write (IOPS).
Through block changes (due to ACTIVE
sessions) in the database, it can be obtained which sessions
are actually consuming the maximum memory and
physical storages which happens only due to DML’s
(INSERT/UPDATE/DELETE/MERGE).
Figure 3: CPU, Memory, I/O and Storage
Fragmentation
Fragmentation in the database occurs mostly due
to DELETE, but also involves
INSERT/UPDATE/MERGE. Fragmentation degrades the
overall database performance, and hence steps needs to be
taken care to remove the fragmentation completely on a
timely basis. By removing the fragmentation, the database
performance in terms of query retrieval is increased and
also the un-used storage is retrieved and which adds up to
the total free space in the database. As the schema size
grows, fragmentation increases, so it’s mandatory that the
task of schedule of fragmentation removal from a schema
needs to be carried out twice a year, so that only the used
space is consumed by the Oracle server, and the unused
space which the Oracle server has, can be reclaimed. DBA
should always see that the schema or tablespace
fragmentation is very minimal so that the DB performance
is maximum.
Index Rebuild
Index rebuild is a continuous on-going activity
which a DBA has to perform in a continuous and timely
phased manner, so that the query retrieval is faster, and the
database performance is increased. Index rebuilds has been
a successfully implemented methodology in the database
technology for the enhanced output of the query, which
has millions of records. Index rebuild is not applicable for
the BLOB’s (Binary Lob’s) like PDF, Image, Videos files,
etc. During the index rebuild activity, it’s very important to
make sure that the schema for which the index is being
rebuild, no sessions of that schema should be there. Also
index rebuild generates archive logs files because Oracle
drops and recreates the index which creates the online redo
log called Archive logs, and, which stores the actual
database changes.
Statistics Gather
Database statistics gathering is an activity which
needs to be performed by the DBA, and which depends
upon the database objects which gets accessed. Statistics
gather for all those objects which have become stale and is
mandatory to be carried out, so that the data dictionary gets
updated, and the updated database details can be retrieved
at any point of time. Statistics gathering is also a very
important factor in the enhancement of the database
performance overall. Statistics gather is also an on-going
activity and which needs to be carried out in a timely
phased manner, which needs to be carried out in the
scheduled DB maintenance time only.
Query Tuning by Using SQL & PL/SQL
SQL is used for faster data retrieval, data
manipulation, transaction control. There are various
statements which relates to SQL such as SELECT,
INSERT, UPDATE, DELETE, MERGE, ALTER, DROP,
COMMIT, TRUNCATE, etc. Many advanced SQL
features have been incorporated in the latest version of
Oracle, and due to this the capability of SQL has increased
a lot.PL/SQL is an extension of SQL with features
incorporating to database programming language, which is
being used widely. Integration, Improved performance,
program development in modules, code clarity and
portability are some of the advantages of PL/SQL. By
using PL/SQL database objects like Procedure, function,
package, package body, trigger, table, views, etc., can be
created easily and which can be used in the application.
Query tuning plays a very important role in the overall
performance of the database.
Below are the Query tuning rules which needs to
be followed in PL/SQL for enhancing the performance of
the database.
 Use ANSI Joins Instead of Oracle Joins.
 Avoid WHERE Clauses with Functions where
ever it is defined.
 Use of CASE Instead of Multiple Unions for
multiple statements in a Query.
 Minimize the Use of DISTINCT clause in every
statement for duplicity of Data.
 Use of UNION ALL instead of UNION in every
statement of result data to be adopted.

 If possible used Global Temporary Table in place
of simple table Creation.
 Always Specify Columns Name in INSERT
Statements.
 Avoid Object Names with “Spaces”.
 Index all the predicates in “WHERE”, “ORDER
BY” and “GROUP BY” clauses.
 Avoid using wild-card (%) at the beginning of a
predicate.
 Avoid unnecessary columns in SELECT clause of
statement.
 Use “inner join”, instead of “outer join” if
possible in statements.
 Tables should always have a Primary Key defined
in columns.
 Do not perform DML statement in Function.
 “As” keyword is used instead of “IS” in
standalone Procedure.
 For error discovery the below command is used
show errors procedure < procedure name>.
 Alternatively, you can type, SHO ERR (short for
SHOW ERRORS) to see the most recent
compilation error.
 “Hard Coded” Values is not be written at
procedural Level. Avoid using it in the Code level
of statements.
 Avoid dynamic creation table at the procedural
level of statements.
 Use of “Exist” and “Not Exists” in place of In
and Not In operator for select statements.
 Avoid “Having Clause” in select statements.
Top 10 database events which are used to define the
overall load on the server, which is shown in the below
graph.
Figure 4: Database load graph AWS RDS
IV. CONCLUSIONS AND
RECOMMENDATIONS
This paper was immensely helpful in
implementing the various features with regards to the
management of the large chain of hotels & ERP
Database world-wide. The widely regarded and
mandatory aspects of the database management in AWS
cloud pertains to the performance of the DB server. Owing
to the growing demand of the database size and taking into
account the core areas of database it’s very much required
to migrate to 19c.By migrating to 19c the company will get
the benefit of Oracle’s ATP (Autonomous Transaction
Processing) feature including Container databases
(CDB/PDB’s) and which provides the HA (High
availability). This is highly appreciated and used globally.
There is vast features in Oracle 19c, and by utilizing
which will increase the overall performance, backup and
recovery & security of the database at large.
As far as recommendations is concerned, we
would like to take this opportunity to implement new
features through the use of latest version of Oracle 19c in
AWS cloud / OCI (Oracle Cloud Infrastructure), which
also supports Autonomous Transaction Processing
(ATP) feature which reduces the manual intervention of
the DBA. Also would like to implement other available
feature in Oracle AWS cloud / OCI Container database
(CDB) and pluggable databases (PDB’s).
Oracle database 19c has new features as
compared to Oracle 12c, and depending upon the current
industry trend it’s the right time to migrate to Oracle
Cloud Infrastructure (OCI), which is a great player in
the database cloud environment management.
REFERENCES
[1] Badger, L., Grance, T., Robert, P., & Voas, J. (2012).
Computer security. DRAFT cloud computing synopsis and
recommendations. NIST Special Publication, pp. 800.
[2] Barbara, J., Jo, A., Cynthia, S., & Adam, T. (2009).
Collaboration using cloud computing and traditional
systems. Western Carolina University, 10(2), 27-32.
[3] Dykstra, J. & Alan, T. (2012). Acquiring evidence
form infrastructure-as-a-service cloud computing:
Exploring the evaluation tools, trust and techniques.
Digital Investigation, S90-S98.
[4] Hofer, C. & Karagiannis, G. (2011). Cloud computing
services: Taxonomy and comparison. Available at:
http://link.springer.com/article/10.1007%2Fs13174-011-
0027-x.
[5] Lazewski, V., Javier, D., Fugang, W., & Geoffrey, C.
(2012). Comparison of multiple cloud frame works.
Available at:

http://www.academia.edu/2929921/Qualitative_Compariso
n_of_Multiple_Cloud_Frameworks.
[6] R. Buyya, C. Vecchiola, & S. T. Selvi. (2013).
Mastering cloud computing: Foundations and applications
programming. Elsevier.
[7] P. K. Paul, K. Kumar, & D. Chaterjee. (2014). Cloud
computing emphasizing emerging possibilities to the entire
information infrastructure. Trends in Information
Management (TRIM), 9(2).
[8] Worldwide cloud computing market shares, strategies,
and forecasts, 2009-2015. (2014 Dec). Available at:
http://www.researchandmarkets.com/reports/1057978/worl
dwide_cloud_computing_market_shares.
[9] P. Mell & T. Grance. (2011). The NIST definition of
Cloud computing. NIST Special Publication.
[10] J. Harris & A. Alter. (2010). Cloud rise: Rewards and
risks at the dawn of cloud computing. Accenture Institute
for High Performance Research Report, pp. 5-30.

Database Performance Management in Cloud

More Related Content

What's hot (20)

Similar to Database Performance Management in Cloud (20)

More from Dr. Amarjeet Singh (20)

Recently uploaded (20)

Database Performance Management in Cloud