SlideShare a Scribd company logo

MySQL 8.0 NF : Common Table Expressions (CTE)

I Goo Lee, profile picture
I Goo Lee

The document discusses common table expressions (CTEs) in MySQL 8.0, highlighting their types such as non-recursive and recursive CTEs, and their advantages over derived tables in terms of readability, performance, and usability. It provides multiple examples demonstrating CTE syntax for SELECT, UPDATE, DELETE, and INSERT operations, as well as how to leverage recursive CTEs for hierarchical data queries. The document also touches on the introduction of descending indexes in MySQL 8.0.

Internet
1 of 44
Downloaded 12 times
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Common Table Expressions By MySQL 8.0 LAB 2017.01.14 (MySQL Power Group) 1
2 Index  Common Table Expression (CTE) 이란?  Non-recursive common table expression  Recursive common table expression  Appendix
3 Common Table Expression  MySQL 8.0 Labs Release 된 New Feature!  기존 파생 테이블은 From절에 subquery 로 표현 SELECT … FROM (subquery) AS drived, t1 …  Common Table Expression(CTE)는 파생테이블과 비슷하지만, From 절 대신 쿼리 블록 앞에 선언! WITH derived AS (subquery) SELECT… FROM derived, t1…  CTE는 subquery가 포함된 SELECT/UPDATE/DELETE에도 사용 가능 WITH derived AS (subquery) DELETE FROM t1 WHERE t1.a IN (SELECT b from derived);
4 EX) SELECT CREATE TABLE t1 ( a INT , b INT ); INSERT INTO t1 VALUES (1,1),(2,2),(3,3); WITH qn AS ( SELECT a+2 AS a, b FROM t1 ) SELECT * FROM qn; Common Table Expression - select
5 Common Table Expression - update EX) UPDATE WITH qn AS ( SELECT a+2 AS a, b FROM t1 ) UPDATE t1, qn SET t1.a=qn.a+10 WHERE t1.a = qn.a; SELECT * FROM t1; self join을 CTE로 작성할 수 있다.
6 Common Table Expression - delete EX) DELETE WITH qn AS ( SELECT a+2 AS a, b FROM t1 ) DELETE t1 FROM t1, qn WHERE t1.a = qn.a; SELECT * FROM t1;
7 Common Table Expression - insert EX) INSERT INSERT INTO t2 WITH qn AS ( SELECT 10*a AS a FROM t1 ) SELECT * FROM qn; SELECT * FROM t2
8 CTE vs 파생테이블 Better readability Can be referenced multiple times Can refer to other CTEs Improved Performance
9 Better readability • 파생테이블 : SELECT * FROM t1 LEFT OUTER JOIN ( SELECT * FROM t2) AS t2 ON t1.id = t2.id LEFT OUTER JOIN ( SELECT * FROM t3) AS t3 ON t1.id = t2.id; • CTE : WITH t2 AS( SELECT * FROM t2 ), t3 AS( SELECT * FROM t3 ) SELECT * FROM t1 LEFT OUTER JOIN t2 ON t1.id = t2.id LEFT OUTER JOIN t2 ON t1.id = t3.id;
10 Can be referenced multiple times • 파생테이블 : SELECT * FROM (SELECT ProductID, COUNT(*) FROM Orders GROUP BY ProductID) A INNER JOIN (SELECT ProductID, COUNT(*) FROM Orders GROUP BY ProductID) B ON A.ProductID = B.ProductID; • CTE : WITH CTE AS ( SELECT ProductID, COUNT(*) AS Cnt FROM Orders GROUP BY ProductID ) SELECT * FROM CTE A INNER JOIN CTE B ON A.ProductID = B.ProductID;
11 Can refer to other CTEs • 파생테이블 : SELECT * FROM (SELECT * FROM Products) A, (SELECT * FROM A) B ERROR : 1146 : Table `A` doesn’t exists • CTE : WITH A AS( SELECT * FROM Prouducts) , B AS( SELECT * FROM A) SELECT * FROM A;  products 테이블은 한번만 참조!
12 Chained CTEs WITH cte1(txt) AS ( SELECT "This " ) , cte2(txt) AS ( SELECT CONCAT(cte1.txt,"is a ") FROM cte1) , cte3(txt) AS ( SELECT "nice query" UNION SELECT "query that rocks" UNION SELECT "query") , cte4(txt) AS ( SELECT concat(cte2.txt, cte3.txt) FROM cte2, cte3 ) SELECT MAX(txt), MIN(txt) FROM cte4; MAX(txt) MIN(txt) This is a query that rocks This a nice query 좋은 예는 아니지만, 이렇게도 쓸 수 있습니다. txt This is a nice query This is a query that rocks This is a query
13 ****테스트 데이터 생성*** -- 1~1000 상품 생성 CREATE TABLE Products SELECT @rownum :=@rownum+1 AS ProductID , CAST(RAND()*100000 AS UNSIGNED) AS Price FROM information_schema.columns A , (SELECT @rownum := 0) B LIMIT 1000; -- 1,000,000건 주문 데이터 생성 CREATE TABLE Orders SELECT DATE_ADD('2010-01-01', INTERVAL CAST(RAND()*3000000 AS UNSIGNED) MINUTE) AS OrderDate , CAST(RAND()*1000 AS UNSIGNED) AS ProductID , CAST(RAND()*100000 AS UNSIGNED) FROM information_schema.columns A , information_schema.columns B , information_schema.columns C LIMIT 1000000; ALTER TABLE Products ADD PRIMARY KEY (ProductID); CREATE INDEX Idx_Orders01 ON Orders (OrderDate); CREATE INDEX Idx_Orders02 ON Orders (ProductID); Improved Performance Products ProductID (PK) Price Orders OrderDate ProductID UserID
14 Improved Performance - view CREATE VIEW v_Orders AS SELECT A.ProductID, SUM(B.Price) AS TotPrice, COUNT(*) AS orderCount FROM Orders A INNER JOIN Products B ON A.ProductID = B.ProductID GROUP BY A.ProductID; SELECT * FROM Products A, v_Orders B WHERE A.ProductID = B.ProductID AND A.ProductID = (SELECT MAX(ProductID) FROM v_Orders); ProductID Price ProductID TotPrice orderCount 1000 55523 1000 30537650 550 1 row in set (1.74 sec)  v_Orders를 두 번 사용할 경우입니다.
15 SubQuery Orders Orders GROUP GROUP SELECT JOIN Products Improved Performance - view
16 WITH CTE AS( SELECT A.ProductID, SUM(B.Price) AS TotPrice, COUNT(*) AS orderCount FROM Orders A INNER JOIN Products B ON A.ProductID = B.ProductID GROUP BY A.ProductID ) SELECT * FROM Products A , CTE B WHERE A.ProductID = B.ProductID AND A.ProductID = (SELECT MAX(ProductID) FROM CTE); Improved Performance - CTE ProductID Price ProductID TotPrice orderCount 1000 55523 1000 30537650 550 1 row in set (0.70 sec)
17 SubQuery OrdersGROUP SELECT JOIN Products SubQuery Improved Performance - CTE
18 Improved Performance 0 0.5 1 1.5 2 View CTE QueryExecutionTime(Seconds) 1.74 sec 0.70 sec
19 Recursive 란? def countdown(n): if n == 0: print "Blastoff!" else: print n countdown(n-1); >>> countdown(3) 3 2 1 Blastoff! 함수가 자기 자신을 호출하는 것! Countdown(n) n == 0 종료! n=n-1
20 Recursive CTE WITH RECURSIVE cte AS( SELECT … FROM table_name /* “seed” SELECT */ UNION ALL SELECT … FROM cte, table_name /* “recursive” SELECT */ ) SELECT … FROM cte; • “seed” SELECT를 한 번 실행하여 초기 데이터 하위 집합을 만들고 재귀 SELECT를 반복 실행하여 전체 결과 집합이 얻어 질 때까지 데이터의 하위 집합을 반환함 • 새로운 Row가 생기지 않을 경우 재귀가 중지됨. • 계층 구조 쿼리를 작성하는데 유용 (부모/자식, 부분/하위)
21 Recursive CTE – Example 1 Print 1 to 10 : WITH RECURSIVE num AS ( SELECT 1 AS n  Seed UNION ALL SELECT 1+n FROM num WHERE n < 10  Recursive ) SELECT n FROM num; n 1 2 3 4 5 6 7 8 9 10
22 Fibonacci Numbers : 0과 1로 시작하며, 다음 피보나치 수는 바로 앞의 두 피보나치 수의 합이 된다. WITH RECURSIVE num AS ( SELECT 1 AS n, 1 AS n1, 1 AS n2 UNION ALL SELECT 1+n, n2, n1+n2 FROM num WHERE n < 10 ) SELECT n, n1, n2 FROM num; Recursive CTE – Example 2 n n1 n2 1 1 1 2 1 2 3 2 3 4 3 5 5 5 8 6 8 13 7 13 21 8 21 34 9 34 55 10 55 89
23 Recursive CTE – Example 3 4-digit bit string print (1/3) WITH RECURSIVE digits AS ( SELECT '0' AS d UNION ALL SELECT '1' ), strings AS ( SELECT CAST('' AS CHAR(4)) AS s UNION ALL SELECT CONCAT(strings.s, digits.d) FROM strings, digits WHERE LENGTH(strings.s) < 4 ) SELECT * FROM strings WHERE LENGTH(s)=4;
24 S 0 1 S 0 1 S 0 0 0 1 1 0 1 1 S 0 1 ………. Recursive CTE – Example 3 해당 구문이 없으면 문자 열 크기가 char(1)로 정해 져 버린다. WITH RECURSIVE digits AS ( SELECT '0' AS d UNION ALL SELECT '1' ), strings AS ( SELECT CAST('' AS CHAR(4)) AS s UNION ALL SELECT CONCAT(strings.s, digits.d) FROM strings, digits WHERE LENGTH(strings.s) < 4 ) SELECT * FROM strings WHERE LENGTH(s)=4; 4-digit bit string print (2/3)
25 Recursive CTE – Example 3 4-digit bit string print (3/3) WITH RECURSIVE digits AS ( SELECT '0' AS d UNION ALL SELECT '1' ), strings AS ( SELECT CAST('' AS CHAR(4)) AS s UNION ALL SELECT CONCAT(strings.s, digits.d) FROM strings, digits WHERE LENGTH(strings.s) < 4 ) SELECT * FROM strings WHERE LENGTH(s)=4;
26 Recursive CTE – Example 4 CREATE TABLE MyEmployees ( EmployeeID VARCHAR(100) NOT NULL, EmployeeName NVARCHAR(30) NOT NULL, ManagerID INT NULL, PRIMARY KEY (EmployeeID ) ); INSERT INTO MyEmployees VALUES (1, N'유재석',NULL) ,(273, N'박명수',1) ,(274, N'정준하',1) ,(275, N'하하',273) ,(276, N'양세형',274) ,(285, N'황광희',273) ,(286, N'정형돈',276) ,(16, N'노홍철', 285) ,(23, N'길', 16); 1 유재석 273 박명수 274 정준하 275 하하 16 노홍철 23 길 285 황광희 276 양세형 286 정형돈
27 Recursive CTE – Example 4 WITH RECURSIVE EmployeeOrder AS ( SELECT EmployeeID, EmployeeName, 1 AS LEVEL, EmployeeID AS Path FROM MyEmployees WHERE ManagerID IS NULL UNION ALL SELECT A.EmployeeID, A.EmployeeName, LEVEL + 1, CONCAT(B.Path, ',', A.EmployeeID) FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.ManagerID = B.EmployeeID ) SELECT * FROM EmployeeOrder; 1. 재귀적 CTE 인 EmployeeOrder는 Seed와 Recursive를 정의함 2. Seed는 직급이 가장 높은 (ManagerID is NULL)로 정의함 3. Level은 Tree의 depth이고, Path는 경로를 나타냄 EmplyeeID EmployeeName Level Path 1 유재석 1 1 [Seed]
28 4. Recursive는 Seed 결과 집합에 있는 직원의 직속 하급자를 반환함 Employee 테이블과 EmployeeOrder CTE 간의 조인작업을 통해 수행되며 첫 번째 반복작업은 다음 결과가 Return 됨 Recursive CTE – Example 4 EmplyeeID EmployeeName Level Path 1 유재석 1 1 273 박명수 2 1,273 274 정준하 2 1,274
29 5. Recursive가 반복적으로 활성화되며, 273, 274를 ManagerID로 가지고 있는 결과를 반환함 EmplyeeID EmployeeName Level Path 1 유재석 1 1 273 박명수 2 1,273 274 정준하 2 1,274 275 하하 3 1,273,275 276 양세형 3 1,274,275 285 황광희 3 1,273,258 Recursive CTE – Example 4
30 WITH RECURSIVE EmployeeOrder AS ( SELECT EmployeeID, EmployeeName, 1 AS LEVEL, EmployeeID AS Path FROM MyEmployees WHERE ManagerID IS NULL UNION ALL SELECT A.EmployeeID, A.EmployeeName, LEVEL + 1, CONCAT(B.Path, ',', A.EmployeeID) FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.ManagerID = B.EmployeeID ) SELECT * FROM EmployeeOrder; Recursive CTE – Example 4 EmployeeID EmployeeName Level Path 1 유재석 1 1 273 박명수 2 1,273 274 정준하 2 1,274 275 하하 3 1,273,275 276 양세형 3 1,274,276 285 황광희 3 1,273,285 16 노홍철 4 1,273,285,16 286 정형돈 4 1,274,276,286 23 길 5 1,273,285,16,23
31 Recursive CTE • 계층적 데이터를 조작 및 구현을 쉽게 할 수 있음  회사 계급 구조  메뉴 구조  가족 관계  주소  게시판 등등… • Recursive CTE에 SELECT 절에는 아래 예약어가 포함될 수 없음  GROUP BY  Aggregate function (SUM, MAX….)  ORDER BY  LIMIT  DISTINCT
32 Recursive CTE – Example 4  특정직원(정준하)의 부하를 찾아보자! WITH RECURSIVE EmployeeOrder AS ( SELECT EmployeeID, EmployeeName, 1 AS LEVEL, EmployeeID AS Path FROM MyEmployees WHERE EmployeeID = 274 (정준하) UNION ALL SELECT A.EmployeeID, A.EmployeeName, LEVEL + 1, CONCAT(B.Path, ',', A.EmployeeID) FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.ManagerID = B.EmployeeID ) SELECT * FROM EmployeeOrder EmplyeeID EmployeeName Level Path 274 정준하 1 274 276 양세형 2 274,276 286 정형돈 3 274.276,286
33 Recursive CTE – Example 4  특정직원(정형돈)의 상사를 찾아보자! WITH RECURSIVE EmployeeOrder AS ( SELECT ManagerID, EmployeeName, 1 AS LEVEL FROM MyEmployees WHERE EmployeeID = 286 (정형돈) UNION ALL SELECT A.ManagerID, A.EmployeeName, 1+LEVEL FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.EmployeeID = B.ManagerID ) SELECT * FROM EmployeeOrder ManagerID EmployeeName Level 276 정형돈 1 274 양세형 2 1 정준하 3 NULL 유재석 4  탐색이 역순이기 때문에 path는 별도 Join으로 나타내어야 함
34 Index  Common Table Expression (CTE) 이란?  Non-recursive common table expression  Recursive common table expression  Appendix # Descending Index
35 Descending Indexes  MySQL 8.0 부터는 Descending Index 지원 !! mysql 5.6> CREATE TABLE t1 (a INT, b INT, INDEX a_desc_b_asc (a DESC, b ASC)); Query OK, 0 rows affected (0.47 sec) mysql 5.6> SHOW CREATE TABLE t1G *************************** 1. row *************************** Table: t1 Create Table: CREATE TABLE `t1` ( `a` int(11) DEFAULT NULL, `b` int(11) DEFAULT NULL, KEY `a_desc_b_asc` (`a`,`b`)  Descending 인덱스로 생성이 되지 않는다. ) ENGINE=InnoDB DEFAULT CHARSET=UTF8MB4 1 row in set (0.00 sec) mysql 8.0> CREATE TABLE t1 (a INT, b INT, INDEX a_desc_b_asc (a DESC, b ASC)); Query OK, 0 rows affected (0.47 sec) mysql 8.0 > SHOW CREATE TABLE t1G *************************** 1. row *************************** Table: t1 Create Table: CREATE TABLE `t1` ( `a` int(11) DEFAULT NULL, `b` int(11) DEFAULT NULL, KEY `a_desc_b_asc` (`a` DESC,`b`) ) ENGINE=InnoDB DEFAULT CHARSET=UTF8MB4 1 row in set (0.00 sec)
36 Descending Indexes Test 환경 (공통) H/W - Amazon EC2 (t2.medium) CPU : 2Core / Memory : 4GB OS - Amazon Linux 비교 DBMS - MySQL 8.0 Labs - MySQL 5.6.29 CREATE TABLE descendingTable ( ID INT DEFAULT NULL, LEVEL INT DEFAULT NULL, KEY (ID DESC, LEVEL ASC) ); INSERT INTO descendingTable SELECT CAST(RAND()*100000 AS UNSIGNED) AS ID , CAST(RAND()*100 AS UNSIGNED) AS LEVEL FROM information_schema.columns A, information_schema.columns B, information_schema.columns C LIMIT 1000000;
37 Descending Indexes - 1 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; Query OK, 1000 rows affected (0.044 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC ID DESC, LEVEL ASC
38 Descending Indexes - 2 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Backward index scan Descending 인덱스가 지원되면서 Backward index scan이라는 연산이 생김 ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
39 Descending Indexes - 3 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.455 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ----------------------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 998412 Using index; Using filesort mysql 8.0> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; • 정렬 조건이 내림/올림 차순이 혼재해 있을 경우 MySQL 5.6버전에서는 Filesort가 발생되며 그 수행속도 차이는 매우 큼 ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
40 Descending Indexes - 4 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.457 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ----------------------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 998412 Using index; Using filesort mysql 8.0> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Backward index scan • MySQL 8.0 에서는 backward index scan을 수행하므로, filesort 없이 빠른 처리가 가능 ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
41 Descending Indexes - 5 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.044 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.452 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Using Filesort • 정의가 되어 있지 않은 정렬을 하므로 MySQL 8.0에서도 filesort가 발생함. 이는 적절한 인덱스를 생성하여 filesort를 방지할 수 있음 ex) (ID DESC, LEVEL DESC) or (ID ASC, LEVEL ASC) ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
42 Descending Indexes - 6 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.457 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Using Filesort ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
43 • MySQL 5.7에서는 query3/4 같은 쿼리의 성능을 향상 시킬 수가 없었으나, MySQL 8.0부터는 Descending Index 를 통해 성능 향상을 할 수 있음 결론
44

More Related Content

PDF
The MySQL Query Optimizer Explained Through Optimizer Trace
oysteing
 
PDF
SWP - A Generic Language Parser
kamaelian
 
KEY
Postgres rules
gisborne
 
DOCX
Clips basics how to make expert system in clips | facts adding | rules makin...
NaumanMalik30
 
PDF
Disassembling Go
Eyal Post
 
PDF
The Ring programming language version 1.5.4 book - Part 22 of 185
Mahmoud Samir Fayed
 
PDF
Functional programming from its fundamentals
Mauro Palsgraaf
 
PPT
Mysql
Yuvaraja Rajenderan
 
The MySQL Query Optimizer Explained Through Optimizer Trace
oysteing
 
SWP - A Generic Language Parser
kamaelian
 
Postgres rules
gisborne
 
Clips basics how to make expert system in clips | facts adding | rules makin...
NaumanMalik30
 
Disassembling Go
Eyal Post
 
The Ring programming language version 1.5.4 book - Part 22 of 185
Mahmoud Samir Fayed
 
Functional programming from its fundamentals
Mauro Palsgraaf
 
Mysql
Yuvaraja Rajenderan
 

What's hot (19)

ODP
Mysql
merlin deepika
 
PPT
SQL querys in detail || Sql query slides
gourav kottawar
 
PDF
Haskell 101
Roberto Pepato
 
PDF
Powerful Explain in MySQL 5.6
MYXPLAIN
 
PDF
MySQL understand Indexes
Roy Yu
 
PPT
SQL || overview and detailed information about Sql
gourav kottawar
 
PPTX
Optimizing queries MySQL
Georgi Sotirov
 
PDF
Go internals (Go Israel Meetup)
Eyal Post
 
PPTX
Pemrograman Terstruktur 4
Moch Mifthachul M
 
ODP
Chetan postgresql partitioning
OpenSourceIndia
 
PDF
MySQL Optimizer: What’s New in 8.0
oysteing
 
PPTX
Sql
ftz 420
 
PDF
MariaDB Optimizer - further down the rabbit hole
Sergey Petrunya
 
PDF
[INSIGHT OUT 2011] B24 effective indexing(tom kyte)
Insight Technology, Inc.
 
PDF
PythonOOP
Veera Pendyala
 
PDF
learn you some erlang - chap0 to chap2
경미 김
 
PPTX
CQL 实现
dennis zhuang
 
PPT
sql statement
zx25 zx25
 
PDF
Python Lab Manual
Bobby Murugesan
 
Mysql
merlin deepika
 
SQL querys in detail || Sql query slides
gourav kottawar
 
Haskell 101
Roberto Pepato
 
Powerful Explain in MySQL 5.6
MYXPLAIN
 
MySQL understand Indexes
Roy Yu
 
SQL || overview and detailed information about Sql
gourav kottawar
 
Optimizing queries MySQL
Georgi Sotirov
 
Go internals (Go Israel Meetup)
Eyal Post
 
Pemrograman Terstruktur 4
Moch Mifthachul M
 
Chetan postgresql partitioning
OpenSourceIndia
 
MySQL Optimizer: What’s New in 8.0
oysteing
 
Sql
ftz 420
 
MariaDB Optimizer - further down the rabbit hole
Sergey Petrunya
 
[INSIGHT OUT 2011] B24 effective indexing(tom kyte)
Insight Technology, Inc.
 
PythonOOP
Veera Pendyala
 
learn you some erlang - chap0 to chap2
경미 김
 
CQL 实现
dennis zhuang
 
sql statement
zx25 zx25
 
Python Lab Manual
Bobby Murugesan
 
Ad

Similar to MySQL 8.0 NF : Common Table Expressions (CTE) (20)

PPT
Mysql
HAINIRMALRAJ
 
PPTX
Set Operators, Derived Tables and CTEs
Aaron Buma
 
PPTX
Set operators - derived tables and CTEs
Steve Stedman
 
PDF
MySQL 8.0: Common Table Expressions
oysteing
 
PDF
MySQL 8.0: Common Table Expressions
oysteing
 
PDF
Table partitioning in PostgreSQL + Rails
Agnieszka Figiel
 
ODP
Mysql1
rajikaa
 
PPTX
Sql
baabtra.com - No. 1 supplier of quality freshers
 
PDF
PPT -The MySQL Query optimizer trace .pdf
ssuserf469dc1
 
PPTX
SQL Data Manipulation language and DQL commands
sonali sonavane
 
PDF
WORKSHEET SQL SOLVED FOR CLASS XII FINAL
RishuKaul2
 
PPTX
Interacting with Oracle Database
Chhom Karath
 
PPT
Oracle tips and tricks
Yanli Liu
 
PDF
Postgres can do THAT?
alexbrasetvik
 
PPSX
Row Pattern Matching in Oracle Database 12c
Stew Ashton
 
PPT
asdasdasdasdsadasdasdasdasdsadasdasdasdsadsadasd
MuhamedAhmed35
 
PDF
dbms lab manual
stalinjothi
 
PPTX
SQL Class Note By Amit Maity PowerPoint Presentation
maitypradip938
 
PPTX
Oraclesql
Priya Goyal
 
PPT
Sql
Karunakaran Mallikeswaran
 
Mysql
HAINIRMALRAJ
 
Set Operators, Derived Tables and CTEs
Aaron Buma
 
Set operators - derived tables and CTEs
Steve Stedman
 
MySQL 8.0: Common Table Expressions
oysteing
 
MySQL 8.0: Common Table Expressions
oysteing
 
Table partitioning in PostgreSQL + Rails
Agnieszka Figiel
 
Mysql1
rajikaa
 
Sql
baabtra.com - No. 1 supplier of quality freshers
 
PPT -The MySQL Query optimizer trace .pdf
ssuserf469dc1
 
SQL Data Manipulation language and DQL commands
sonali sonavane
 
WORKSHEET SQL SOLVED FOR CLASS XII FINAL
RishuKaul2
 
Interacting with Oracle Database
Chhom Karath
 
Oracle tips and tricks
Yanli Liu
 
Postgres can do THAT?
alexbrasetvik
 
Row Pattern Matching in Oracle Database 12c
Stew Ashton
 
asdasdasdasdsadasdasdasdasdsadasdasdasdsadsadasd
MuhamedAhmed35
 
dbms lab manual
stalinjothi
 
SQL Class Note By Amit Maity PowerPoint Presentation
maitypradip938
 
Oraclesql
Priya Goyal
 
Sql
Karunakaran Mallikeswaran
 
Ad

More from I Goo Lee (20)

PDF
MySQL_Fabric_운영시유의사항
I Goo Lee
 
PDF
MySQL Deep dive with FusionIO
I Goo Lee
 
PDF
From MSSQL to MySQL
I Goo Lee
 
PDF
From MSSQL to MariaDB
I Goo Lee
 
PDF
AWS Aurora 100% 활용하기
I Goo Lee
 
PDF
Backup automation in KAKAO
I Goo Lee
 
PDF
텔레그램을 이용한 양방향 모니터링 시스템 구축
I Goo Lee
 
PDF
Federated Engine 실무적용사례
I Goo Lee
 
PDF
MySQL 상태 메시지 분석 및 활용
I Goo Lee
 
PDF
MySQL 5.7 NF – Optimizer Improvement
I Goo Lee
 
PDF
MySQL 5.7 NF – JSON Datatype 활용
I Goo Lee
 
PDF
Intro KaKao MRTE (MySQL Realtime Traffic Emulator)
I Goo Lee
 
PDF
MS 빅데이터 서비스 및 게임사 PoC 사례 소개
I Goo Lee
 
PDF
AWS 환경에서 MySQL Infra 설계하기-2본론
I Goo Lee
 
PDF
AWS 환경에서 MySQL Infra 설계하기-1도입부분
I Goo Lee
 
PDF
AWS 환경에서 MySQL BMT
I Goo Lee
 
PDF
MySQL Slow Query log Monitoring using Beats & ELK
I Goo Lee
 
PDF
MySQL Audit using Percona audit plugin and ELK
I Goo Lee
 
PDF
PostgreSQL 이야기
I Goo Lee
 
PDF
Intro KaKao ADT (Almighty Data Transmitter)
I Goo Lee
 
MySQL_Fabric_운영시유의사항
I Goo Lee
 
MySQL Deep dive with FusionIO
I Goo Lee
 
From MSSQL to MySQL
I Goo Lee
 
From MSSQL to MariaDB
I Goo Lee
 
AWS Aurora 100% 활용하기
I Goo Lee
 
Backup automation in KAKAO
I Goo Lee
 
텔레그램을 이용한 양방향 모니터링 시스템 구축
I Goo Lee
 
Federated Engine 실무적용사례
I Goo Lee
 
MySQL 상태 메시지 분석 및 활용
I Goo Lee
 
MySQL 5.7 NF – Optimizer Improvement
I Goo Lee
 
MySQL 5.7 NF – JSON Datatype 활용
I Goo Lee
 
Intro KaKao MRTE (MySQL Realtime Traffic Emulator)
I Goo Lee
 
MS 빅데이터 서비스 및 게임사 PoC 사례 소개
I Goo Lee
 
AWS 환경에서 MySQL Infra 설계하기-2본론
I Goo Lee
 
AWS 환경에서 MySQL Infra 설계하기-1도입부분
I Goo Lee
 
AWS 환경에서 MySQL BMT
I Goo Lee
 
MySQL Slow Query log Monitoring using Beats & ELK
I Goo Lee
 
MySQL Audit using Percona audit plugin and ELK
I Goo Lee
 
PostgreSQL 이야기
I Goo Lee
 
Intro KaKao ADT (Almighty Data Transmitter)
I Goo Lee
 

Recently uploaded (20)

PPT
tcp ip networks nd ip layering assotred slides
pakadamfdp
 
PDF
Best Practices for Testing and Debugging Shopify Third-Party API Integrations...
CartCoders
 
PDF
💰 𝐔𝐊𝐓𝐈 𝐊𝐄𝐌𝐄𝐍𝐀𝐍𝐆𝐀𝐍 𝐊𝐈𝐏𝐄𝐑𝟒𝐃 𝐇𝐀𝐑𝐈 𝐈𝐍𝐈 𝟐𝟎𝟐𝟓 💰
GRAB
 
PPTX
Slides PPTX World Game (s) Eco Economic Epochs.pptx
Steven McGee
 
PDF
Cloud-Scale Log Monitoring _ Datadog.pdf
MuhammadDhomanhuriMa
 
PPTX
artificial intelligence overview of it and more
yafotin445
 
PPTX
Introuction about ICD -10 and ICD-11 PPT.pptx
naveenithkrishnan
 
PPTX
522797556-Unit-2-Temperature-measurement-1-1.pptx
msar8888
 
PDF
Tenda Login Guide: Access Your Router in 5 Easy Steps
tendawifi16
 
PDF
Exploring VPS Hosting Trends for SMBs in 2025
Liquid Web
 
PPTX
Introduction to Information and Communication Technology
AkmadArzieAyao1
 
PDF
The New Creative Director: How AI Tools for Social Media Content Creation Are...
SageTitans
 
DOCX
Unit-3 cyber security network security of internet system
shivangisharma636228
 
PDF
An introduction to the IFRS (ISSB) Stndards.pdf
farhankhan13694
 
PDF
Vigrab.top – Online Tool for Downloading and Converting Social Media Videos a...
Vigrab Top
 
PDF
Introduction to the IoT system, how the IoT system works
TinBinh
 
PPT
Ethics in Information System - Management Information System
faizhossain3
 
PDF
Paper PDF World Game (s) Great Redesign.pdf
Steven McGee
 
PPTX
Power Point - Lesson 3_2.pptx grad school presentation
JoyPPD
 
PPTX
artificialintelligenceai1-copy-210604123353.pptx
b45r14
 
tcp ip networks nd ip layering assotred slides
pakadamfdp
 
Best Practices for Testing and Debugging Shopify Third-Party API Integrations...
CartCoders
 
💰 𝐔𝐊𝐓𝐈 𝐊𝐄𝐌𝐄𝐍𝐀𝐍𝐆𝐀𝐍 𝐊𝐈𝐏𝐄𝐑𝟒𝐃 𝐇𝐀𝐑𝐈 𝐈𝐍𝐈 𝟐𝟎𝟐𝟓 💰
GRAB
 
Slides PPTX World Game (s) Eco Economic Epochs.pptx
Steven McGee
 
Cloud-Scale Log Monitoring _ Datadog.pdf
MuhammadDhomanhuriMa
 
artificial intelligence overview of it and more
yafotin445
 
Introuction about ICD -10 and ICD-11 PPT.pptx
naveenithkrishnan
 
522797556-Unit-2-Temperature-measurement-1-1.pptx
msar8888
 
Tenda Login Guide: Access Your Router in 5 Easy Steps
tendawifi16
 
Exploring VPS Hosting Trends for SMBs in 2025
Liquid Web
 
Introduction to Information and Communication Technology
AkmadArzieAyao1
 
The New Creative Director: How AI Tools for Social Media Content Creation Are...
SageTitans
 
Unit-3 cyber security network security of internet system
shivangisharma636228
 
An introduction to the IFRS (ISSB) Stndards.pdf
farhankhan13694
 
Vigrab.top – Online Tool for Downloading and Converting Social Media Videos a...
Vigrab Top
 
Introduction to the IoT system, how the IoT system works
TinBinh
 
Ethics in Information System - Management Information System
faizhossain3
 
Paper PDF World Game (s) Great Redesign.pdf
Steven McGee
 
Power Point - Lesson 3_2.pptx grad school presentation
JoyPPD
 
artificialintelligenceai1-copy-210604123353.pptx
b45r14
 

MySQL 8.0 NF : Common Table Expressions (CTE)

  • 1. Common Table Expressions By MySQL 8.0 LAB 2017.01.14 (MySQL Power Group) 1
  • 2. 2 Index  Common Table Expression (CTE) 이란?  Non-recursive common table expression  Recursive common table expression  Appendix
  • 3. 3 Common Table Expression  MySQL 8.0 Labs Release 된 New Feature!  기존 파생 테이블은 From절에 subquery 로 표현 SELECT … FROM (subquery) AS drived, t1 …  Common Table Expression(CTE)는 파생테이블과 비슷하지만, From 절 대신 쿼리 블록 앞에 선언! WITH derived AS (subquery) SELECT… FROM derived, t1…  CTE는 subquery가 포함된 SELECT/UPDATE/DELETE에도 사용 가능 WITH derived AS (subquery) DELETE FROM t1 WHERE t1.a IN (SELECT b from derived);
  • 4. 4 EX) SELECT CREATE TABLE t1 ( a INT , b INT ); INSERT INTO t1 VALUES (1,1),(2,2),(3,3); WITH qn AS ( SELECT a+2 AS a, b FROM t1 ) SELECT * FROM qn; Common Table Expression - select
  • 5. 5 Common Table Expression - update EX) UPDATE WITH qn AS ( SELECT a+2 AS a, b FROM t1 ) UPDATE t1, qn SET t1.a=qn.a+10 WHERE t1.a = qn.a; SELECT * FROM t1; self join을 CTE로 작성할 수 있다.
  • 6. 6 Common Table Expression - delete EX) DELETE WITH qn AS ( SELECT a+2 AS a, b FROM t1 ) DELETE t1 FROM t1, qn WHERE t1.a = qn.a; SELECT * FROM t1;
  • 7. 7 Common Table Expression - insert EX) INSERT INSERT INTO t2 WITH qn AS ( SELECT 10*a AS a FROM t1 ) SELECT * FROM qn; SELECT * FROM t2
  • 8. 8 CTE vs 파생테이블 Better readability Can be referenced multiple times Can refer to other CTEs Improved Performance
  • 9. 9 Better readability • 파생테이블 : SELECT * FROM t1 LEFT OUTER JOIN ( SELECT * FROM t2) AS t2 ON t1.id = t2.id LEFT OUTER JOIN ( SELECT * FROM t3) AS t3 ON t1.id = t2.id; • CTE : WITH t2 AS( SELECT * FROM t2 ), t3 AS( SELECT * FROM t3 ) SELECT * FROM t1 LEFT OUTER JOIN t2 ON t1.id = t2.id LEFT OUTER JOIN t2 ON t1.id = t3.id;
  • 10. 10 Can be referenced multiple times • 파생테이블 : SELECT * FROM (SELECT ProductID, COUNT(*) FROM Orders GROUP BY ProductID) A INNER JOIN (SELECT ProductID, COUNT(*) FROM Orders GROUP BY ProductID) B ON A.ProductID = B.ProductID; • CTE : WITH CTE AS ( SELECT ProductID, COUNT(*) AS Cnt FROM Orders GROUP BY ProductID ) SELECT * FROM CTE A INNER JOIN CTE B ON A.ProductID = B.ProductID;
  • 11. 11 Can refer to other CTEs • 파생테이블 : SELECT * FROM (SELECT * FROM Products) A, (SELECT * FROM A) B ERROR : 1146 : Table `A` doesn’t exists • CTE : WITH A AS( SELECT * FROM Prouducts) , B AS( SELECT * FROM A) SELECT * FROM A;  products 테이블은 한번만 참조!
  • 12. 12 Chained CTEs WITH cte1(txt) AS ( SELECT "This " ) , cte2(txt) AS ( SELECT CONCAT(cte1.txt,"is a ") FROM cte1) , cte3(txt) AS ( SELECT "nice query" UNION SELECT "query that rocks" UNION SELECT "query") , cte4(txt) AS ( SELECT concat(cte2.txt, cte3.txt) FROM cte2, cte3 ) SELECT MAX(txt), MIN(txt) FROM cte4; MAX(txt) MIN(txt) This is a query that rocks This a nice query 좋은 예는 아니지만, 이렇게도 쓸 수 있습니다. txt This is a nice query This is a query that rocks This is a query
  • 13. 13 ****테스트 데이터 생성*** -- 1~1000 상품 생성 CREATE TABLE Products SELECT @rownum :=@rownum+1 AS ProductID , CAST(RAND()*100000 AS UNSIGNED) AS Price FROM information_schema.columns A , (SELECT @rownum := 0) B LIMIT 1000; -- 1,000,000건 주문 데이터 생성 CREATE TABLE Orders SELECT DATE_ADD('2010-01-01', INTERVAL CAST(RAND()*3000000 AS UNSIGNED) MINUTE) AS OrderDate , CAST(RAND()*1000 AS UNSIGNED) AS ProductID , CAST(RAND()*100000 AS UNSIGNED) FROM information_schema.columns A , information_schema.columns B , information_schema.columns C LIMIT 1000000; ALTER TABLE Products ADD PRIMARY KEY (ProductID); CREATE INDEX Idx_Orders01 ON Orders (OrderDate); CREATE INDEX Idx_Orders02 ON Orders (ProductID); Improved Performance Products ProductID (PK) Price Orders OrderDate ProductID UserID
  • 14. 14 Improved Performance - view CREATE VIEW v_Orders AS SELECT A.ProductID, SUM(B.Price) AS TotPrice, COUNT(*) AS orderCount FROM Orders A INNER JOIN Products B ON A.ProductID = B.ProductID GROUP BY A.ProductID; SELECT * FROM Products A, v_Orders B WHERE A.ProductID = B.ProductID AND A.ProductID = (SELECT MAX(ProductID) FROM v_Orders); ProductID Price ProductID TotPrice orderCount 1000 55523 1000 30537650 550 1 row in set (1.74 sec)  v_Orders를 두 번 사용할 경우입니다.
  • 16. 16 WITH CTE AS( SELECT A.ProductID, SUM(B.Price) AS TotPrice, COUNT(*) AS orderCount FROM Orders A INNER JOIN Products B ON A.ProductID = B.ProductID GROUP BY A.ProductID ) SELECT * FROM Products A , CTE B WHERE A.ProductID = B.ProductID AND A.ProductID = (SELECT MAX(ProductID) FROM CTE); Improved Performance - CTE ProductID Price ProductID TotPrice orderCount 1000 55523 1000 30537650 550 1 row in set (0.70 sec)
  • 19. 19 Recursive 란? def countdown(n): if n == 0: print "Blastoff!" else: print n countdown(n-1); >>> countdown(3) 3 2 1 Blastoff! 함수가 자기 자신을 호출하는 것! Countdown(n) n == 0 종료! n=n-1
  • 20. 20 Recursive CTE WITH RECURSIVE cte AS( SELECT … FROM table_name /* “seed” SELECT */ UNION ALL SELECT … FROM cte, table_name /* “recursive” SELECT */ ) SELECT … FROM cte; • “seed” SELECT를 한 번 실행하여 초기 데이터 하위 집합을 만들고 재귀 SELECT를 반복 실행하여 전체 결과 집합이 얻어 질 때까지 데이터의 하위 집합을 반환함 • 새로운 Row가 생기지 않을 경우 재귀가 중지됨. • 계층 구조 쿼리를 작성하는데 유용 (부모/자식, 부분/하위)
  • 21. 21 Recursive CTE – Example 1 Print 1 to 10 : WITH RECURSIVE num AS ( SELECT 1 AS n  Seed UNION ALL SELECT 1+n FROM num WHERE n < 10  Recursive ) SELECT n FROM num; n 1 2 3 4 5 6 7 8 9 10
  • 22. 22 Fibonacci Numbers : 0과 1로 시작하며, 다음 피보나치 수는 바로 앞의 두 피보나치 수의 합이 된다. WITH RECURSIVE num AS ( SELECT 1 AS n, 1 AS n1, 1 AS n2 UNION ALL SELECT 1+n, n2, n1+n2 FROM num WHERE n < 10 ) SELECT n, n1, n2 FROM num; Recursive CTE – Example 2 n n1 n2 1 1 1 2 1 2 3 2 3 4 3 5 5 5 8 6 8 13 7 13 21 8 21 34 9 34 55 10 55 89
  • 23. 23 Recursive CTE – Example 3 4-digit bit string print (1/3) WITH RECURSIVE digits AS ( SELECT '0' AS d UNION ALL SELECT '1' ), strings AS ( SELECT CAST('' AS CHAR(4)) AS s UNION ALL SELECT CONCAT(strings.s, digits.d) FROM strings, digits WHERE LENGTH(strings.s) < 4 ) SELECT * FROM strings WHERE LENGTH(s)=4;
  • 24. 24 S 0 1 S 0 1 S 0 0 0 1 1 0 1 1 S 0 1 ………. Recursive CTE – Example 3 해당 구문이 없으면 문자 열 크기가 char(1)로 정해 져 버린다. WITH RECURSIVE digits AS ( SELECT '0' AS d UNION ALL SELECT '1' ), strings AS ( SELECT CAST('' AS CHAR(4)) AS s UNION ALL SELECT CONCAT(strings.s, digits.d) FROM strings, digits WHERE LENGTH(strings.s) < 4 ) SELECT * FROM strings WHERE LENGTH(s)=4; 4-digit bit string print (2/3)
  • 25. 25 Recursive CTE – Example 3 4-digit bit string print (3/3) WITH RECURSIVE digits AS ( SELECT '0' AS d UNION ALL SELECT '1' ), strings AS ( SELECT CAST('' AS CHAR(4)) AS s UNION ALL SELECT CONCAT(strings.s, digits.d) FROM strings, digits WHERE LENGTH(strings.s) < 4 ) SELECT * FROM strings WHERE LENGTH(s)=4;
  • 26. 26 Recursive CTE – Example 4 CREATE TABLE MyEmployees ( EmployeeID VARCHAR(100) NOT NULL, EmployeeName NVARCHAR(30) NOT NULL, ManagerID INT NULL, PRIMARY KEY (EmployeeID ) ); INSERT INTO MyEmployees VALUES (1, N'유재석',NULL) ,(273, N'박명수',1) ,(274, N'정준하',1) ,(275, N'하하',273) ,(276, N'양세형',274) ,(285, N'황광희',273) ,(286, N'정형돈',276) ,(16, N'노홍철', 285) ,(23, N'길', 16); 1 유재석 273 박명수 274 정준하 275 하하 16 노홍철 23 길 285 황광희 276 양세형 286 정형돈
  • 27. 27 Recursive CTE – Example 4 WITH RECURSIVE EmployeeOrder AS ( SELECT EmployeeID, EmployeeName, 1 AS LEVEL, EmployeeID AS Path FROM MyEmployees WHERE ManagerID IS NULL UNION ALL SELECT A.EmployeeID, A.EmployeeName, LEVEL + 1, CONCAT(B.Path, ',', A.EmployeeID) FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.ManagerID = B.EmployeeID ) SELECT * FROM EmployeeOrder; 1. 재귀적 CTE 인 EmployeeOrder는 Seed와 Recursive를 정의함 2. Seed는 직급이 가장 높은 (ManagerID is NULL)로 정의함 3. Level은 Tree의 depth이고, Path는 경로를 나타냄 EmplyeeID EmployeeName Level Path 1 유재석 1 1 [Seed]
  • 28. 28 4. Recursive는 Seed 결과 집합에 있는 직원의 직속 하급자를 반환함 Employee 테이블과 EmployeeOrder CTE 간의 조인작업을 통해 수행되며 첫 번째 반복작업은 다음 결과가 Return 됨 Recursive CTE – Example 4 EmplyeeID EmployeeName Level Path 1 유재석 1 1 273 박명수 2 1,273 274 정준하 2 1,274
  • 29. 29 5. Recursive가 반복적으로 활성화되며, 273, 274를 ManagerID로 가지고 있는 결과를 반환함 EmplyeeID EmployeeName Level Path 1 유재석 1 1 273 박명수 2 1,273 274 정준하 2 1,274 275 하하 3 1,273,275 276 양세형 3 1,274,275 285 황광희 3 1,273,258 Recursive CTE – Example 4
  • 30. 30 WITH RECURSIVE EmployeeOrder AS ( SELECT EmployeeID, EmployeeName, 1 AS LEVEL, EmployeeID AS Path FROM MyEmployees WHERE ManagerID IS NULL UNION ALL SELECT A.EmployeeID, A.EmployeeName, LEVEL + 1, CONCAT(B.Path, ',', A.EmployeeID) FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.ManagerID = B.EmployeeID ) SELECT * FROM EmployeeOrder; Recursive CTE – Example 4 EmployeeID EmployeeName Level Path 1 유재석 1 1 273 박명수 2 1,273 274 정준하 2 1,274 275 하하 3 1,273,275 276 양세형 3 1,274,276 285 황광희 3 1,273,285 16 노홍철 4 1,273,285,16 286 정형돈 4 1,274,276,286 23 길 5 1,273,285,16,23
  • 31. 31 Recursive CTE • 계층적 데이터를 조작 및 구현을 쉽게 할 수 있음  회사 계급 구조  메뉴 구조  가족 관계  주소  게시판 등등… • Recursive CTE에 SELECT 절에는 아래 예약어가 포함될 수 없음  GROUP BY  Aggregate function (SUM, MAX….)  ORDER BY  LIMIT  DISTINCT
  • 32. 32 Recursive CTE – Example 4  특정직원(정준하)의 부하를 찾아보자! WITH RECURSIVE EmployeeOrder AS ( SELECT EmployeeID, EmployeeName, 1 AS LEVEL, EmployeeID AS Path FROM MyEmployees WHERE EmployeeID = 274 (정준하) UNION ALL SELECT A.EmployeeID, A.EmployeeName, LEVEL + 1, CONCAT(B.Path, ',', A.EmployeeID) FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.ManagerID = B.EmployeeID ) SELECT * FROM EmployeeOrder EmplyeeID EmployeeName Level Path 274 정준하 1 274 276 양세형 2 274,276 286 정형돈 3 274.276,286
  • 33. 33 Recursive CTE – Example 4  특정직원(정형돈)의 상사를 찾아보자! WITH RECURSIVE EmployeeOrder AS ( SELECT ManagerID, EmployeeName, 1 AS LEVEL FROM MyEmployees WHERE EmployeeID = 286 (정형돈) UNION ALL SELECT A.ManagerID, A.EmployeeName, 1+LEVEL FROM MyEmployees A INNER JOIN EmployeeOrder B ON A.EmployeeID = B.ManagerID ) SELECT * FROM EmployeeOrder ManagerID EmployeeName Level 276 정형돈 1 274 양세형 2 1 정준하 3 NULL 유재석 4  탐색이 역순이기 때문에 path는 별도 Join으로 나타내어야 함
  • 34. 34 Index  Common Table Expression (CTE) 이란?  Non-recursive common table expression  Recursive common table expression  Appendix # Descending Index
  • 35. 35 Descending Indexes  MySQL 8.0 부터는 Descending Index 지원 !! mysql 5.6> CREATE TABLE t1 (a INT, b INT, INDEX a_desc_b_asc (a DESC, b ASC)); Query OK, 0 rows affected (0.47 sec) mysql 5.6> SHOW CREATE TABLE t1G *************************** 1. row *************************** Table: t1 Create Table: CREATE TABLE `t1` ( `a` int(11) DEFAULT NULL, `b` int(11) DEFAULT NULL, KEY `a_desc_b_asc` (`a`,`b`)  Descending 인덱스로 생성이 되지 않는다. ) ENGINE=InnoDB DEFAULT CHARSET=UTF8MB4 1 row in set (0.00 sec) mysql 8.0> CREATE TABLE t1 (a INT, b INT, INDEX a_desc_b_asc (a DESC, b ASC)); Query OK, 0 rows affected (0.47 sec) mysql 8.0 > SHOW CREATE TABLE t1G *************************** 1. row *************************** Table: t1 Create Table: CREATE TABLE `t1` ( `a` int(11) DEFAULT NULL, `b` int(11) DEFAULT NULL, KEY `a_desc_b_asc` (`a` DESC,`b`) ) ENGINE=InnoDB DEFAULT CHARSET=UTF8MB4 1 row in set (0.00 sec)
  • 36. 36 Descending Indexes Test 환경 (공통) H/W - Amazon EC2 (t2.medium) CPU : 2Core / Memory : 4GB OS - Amazon Linux 비교 DBMS - MySQL 8.0 Labs - MySQL 5.6.29 CREATE TABLE descendingTable ( ID INT DEFAULT NULL, LEVEL INT DEFAULT NULL, KEY (ID DESC, LEVEL ASC) ); INSERT INTO descendingTable SELECT CAST(RAND()*100000 AS UNSIGNED) AS ID , CAST(RAND()*100 AS UNSIGNED) AS LEVEL FROM information_schema.columns A, information_schema.columns B, information_schema.columns C LIMIT 1000000;
  • 37. 37 Descending Indexes - 1 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; Query OK, 1000 rows affected (0.044 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC ID DESC, LEVEL ASC
  • 38. 38 Descending Indexes - 2 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Backward index scan Descending 인덱스가 지원되면서 Backward index scan이라는 연산이 생김 ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
  • 39. 39 Descending Indexes - 3 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.455 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ----------------------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 998412 Using index; Using filesort mysql 8.0> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; • 정렬 조건이 내림/올림 차순이 혼재해 있을 경우 MySQL 5.6버전에서는 Filesort가 발생되며 그 수행속도 차이는 매우 큼 ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
  • 40. 40 Descending Indexes - 4 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.457 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ----------------------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 998412 Using index; Using filesort mysql 8.0> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL DESC LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Backward index scan • MySQL 8.0 에서는 backward index scan을 수행하므로, filesort 없이 빠른 처리가 가능 ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
  • 41. 41 Descending Indexes - 5 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.044 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; Query OK, 1000 rows affected (0.452 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID DESC, LEVEL DESC LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Using Filesort • 정의가 되어 있지 않은 정렬을 하므로 MySQL 8.0에서도 filesort가 발생함. 이는 적절한 인덱스를 생성하여 filesort를 방지할 수 있음 ex) (ID DESC, LEVEL DESC) or (ID ASC, LEVEL ASC) ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
  • 42. 42 Descending Indexes - 6 mysql 5.6> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.457 sec) mysql 5.6> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; id select_type table type possible_keys key key_len ref rows Extra ------ ----------- --------------- ------ ------------- ------ ------- ------ ------ ------------- 1 SIMPLE descendingTable index (NULL) ID 10 (NULL) 1000 Using index mysql 8.0> SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; Query OK, 1000 rows affected (0.042 sec) mysql 8.0> EXPLAIN SELECT * FROM descendingTable ORDER BY ID ASC, LEVEL ASC LIMIT 1000; partitions type possible_keys key key_len ref rows filtered Extra ---------- ------ ------------- ------ ------- ------ ------ -------- ---------------------------------- (NULL) index (NULL) ID 10 (NULL) 1000 100.00 Using index; Using Filesort ID DESC, LEVEL ASC ID DESC, LEVEL ASC ⇒ ID ASC, LEVEL ASC
  • 43. 43 • MySQL 5.7에서는 query3/4 같은 쿼리의 성능을 향상 시킬 수가 없었으나, MySQL 8.0부터는 Descending Index 를 통해 성능 향상을 할 수 있음 결론
  • 44. 44