IDUG 2013 Sheryl Larsen Tuning SQL

1

z/OS

Application Development

Tuning SQL for Performance

Sheryl M. Larsen Sheryl M. Larsen Sheryl M. Larsen, Inc.Sheryl M. Larsen, Inc.

Platform:

2013

IDUG, World Wide

2© Sheryl M. Larsen, Inc. 2000-2011Contact: [email protected] (630) 399-3330 WWW.SMLSQL.COMContact: [email protected] (630) 399-3330 WWW.SMLSQL.COM

Sheryl M. Larsen, Inc.DB2 SQL Consulting & Education

Sheryl Larsen is an internationally recognized researcher, consultant and lecturer, specializing in DB2 and is known for her extensive expertise in SQL. She co-authored a book, DB2 Answers, Osborne-McGraw-Hill, 1999.

She was voted an inaugural IBM Information Champion 2009- 2011, IDUG Speaker Hall of Fame and, a member of IBM’s DB2 Gold Consultants program since 1994.

Currently, she is President of the Midwest Database Users Group (mwdug.org), and owns Sheryl M. Larsen, Inc., a firm specializing in Advanced DB2 Consulting and Education.

Sheryl has 25 years experience in DB2, has published articles, white papers, webtv, animated

access paths:

Google Sheryl Larsen @Bmc.com Ca.com Ibm.com

3© Sheryl M. Larsen, Inc. 2000-2011

SQL PerformanceDB2 Engine ComponentsPredicate Processing IntelligenceOptimizer DetailsSQL Performance Rules

Tuning QueriesWhen? Why? How? Introduction to Proven SQL Tuning Methods

When are Access Paths Good or Bad? Variations of index accessVariations of table accessVariations of join methods

Reading the Optimizer’s MindPLAN_TABLE OutputDSN_STATEMENT Output

Table of Contents - TuningCase Studies Using a Proven Method

Tuning Example 1 – OPTIMIZE FOR n ROWS/FETCH FIRST n ROWS ONLYTuning Example 2, 3, 4 – No OperationsTuning Example 5, 6 – Fake FilteringTuning Example 7 – Index Design

Extreme TuningTuning Example 8 – Distinct Table Expressions Tuning Example 9 – Anti-JoinsTuning Example 10 – (Predicate OR 0 = 1)Tuning Example 11 – Extreme Cross Query Block Optimization

Designing Performance Structures Index Design for a Single QueryIndex Design for a WorkloadMQT Design for a Workload

4

SQL Performance

DB2 Engine ComponentsDB2 Engine Components

Predicate Processing IntelligencePredicate Processing Intelligence

Optimizer DetailsOptimizer Details

SQL Performance RulesSQL Performance Rules

5

DirectoryDirectory

WorkWorkFilesFiles

Buffer Pool

SQLSQL

CatalogCatalog

DataData

Buffer ManagerBuffer Manager

ResultResult

DB2 Engine Components

Stage 2 – z/OSStage 2 – z/OSResidual – LUW Residual – LUW

IndexIndex

Stage 1 – z/OSStage 1 – z/OSSargable – LUW Sargable – LUW

SQL Execution

DirectoryDirectory

CatalogCatalog

Access Plans

Meta Data

Optimizer

RIDPool

RIDs

Dynamic Statement Cache

DB2 V8 Enhancements

6

DirectoryDirectory

WorkWorkFilesFiles

Buffer Pool

SQL

CatalogCatalog

DataData

Buffer Manager

Result

Stage 2

IndexIndex

Stage 1 DirectoryDirectory

CatalogCatalog

8K 16K 32K Pages

IS NOT NULLIS NOT NULLIS NOT DISTINCT FROMIS NOT DISTINCT FROM

Index only on VARCHARIndexes

Top & Bottom NUDS

Materialized Query Tables

2M Statement Length Multi-row FETCHMulti-row INSERT

SELECT INTO w/ ORDER BYCommon Table Expressions

Recursive SQLIS NOT DISCTINCT FROM

Scalar FullselectGROUP BY Expression

Dynamic Scrollable CursorsINSERT with SELECT

Multiple DISTINCT ClausesSequences

Parallel SortParallel Sort

Forward and Backward IndexesDPSI Indexes

Stage 1 Unlike Data TypesStage 1 Unlike Data Types

64 Bit Addressing

More Hybrid Joins

© SMLI 2009

Access Plans

Meta Data

Optimizer

RIDPool

RIDs

Dynamic Statement

Cache

DB2 9 – WOW!


DirectoryDirectory

WorkWorkFilesFiles

Buffer Pool

SQL

CatalogCatalog

DataData

Buffer Manager

Result

Stage 2

IndexIndex

Stage 1 DirectoryDirectory

CatalogCatalog

In MemoryGLOBAL TEMP TABLES

Views Table Expressions

In Memory

Faster Sorts for GROUP BYDISTINCT, ORDER BY

Index on Expression

Histogram Statistics

Page Range Index ScreeningFor Partitioned Indexes

Global Optimization

TRUNCATEMERGE

SELECT FROM UPDATE/DELETEEXCEPT, INTERSECTOLAP Expressions

ORDER BY / FETCH FIRST in Subselect

More Parallelism

Sparse Indexesfor more Work Files

Access Plans

Meta Data

Optimizer

RIDPool

RIDs

Dynamic Statement

Cache


4K

WorkWorkFilesFiles

4K Page Processing – z/OS

Buffer Manager

DataData IndexIndex

Buffer Pool

Stage 2

Stage 1


Indexable Stage 1 Predicates

Predicate Type Indexable Stage 1

COL = value Y Y COL = noncol expr Y Y COL IS NULL Y Y COL op value Y Y COL op noncol expr Y Y COL BETWEEN value1 AND value2

Y Y

COL BETWEEN noncol expr1 AND noncol expr2

Y Y

COL LIKE 'pattern' Y Y COL IN (list) Y Y COL LIKE host variable Y Y T1.COL = T2.COL Y Y T1.COL op T2.COL Y Y COL=(non subq) Y Y COL op (non subq) Y Y COL op ANY (non subq) Y Y COL op ALL (non subq) Y Y COL IN (non subq) Y Y COL = expression Y Y (COL1,...COLn) IN (non subq) Y Y (COL1, …COLn) = (value1, …valuen)

Y Y

T1.COL = T2.colexpr Y Y COL IS NOT NULL Y Y COL IS NOT DISTINCT FROM value

Y Y

COL IS NOT DISTINCT FROM noncol expression

Y Y

COL IS NOT DISTINCT FROM col expression

Y Y

COL IS NOT DISTINCT FROM non subq

Y Y

T1.COL IS NOT DISTINCT FROM T2.COL

Y Y

T1.COL IS NOT DISTINCT FROM T2.col expression

Y Y

Stage 1 Predicates

Predicate Type

Indexable Stage 1

COL <> value N Y COL <> noncol expr N Y COL NOT BETWEEN value1 AND value2

N Y

COL NOT BETWEEN noncol expr1 AND noncol expr2

N Y

COL NOT IN (list) N Y COL NOT LIKE ' char' N Y COL LIKE '%char' N Y COL LIKE '_char' N Y T1.COL <> T2.COL N Y T1.COL1 = T1.COL2 N Y COL <> (non subq) N Y COL IS DISTINCT FROM N Y

1. Indexable = The predicate is a candidate for Matching Index access. When the optimizer chooses to use a predicate in the probe of the index, the condition is named Matching (matching the index). This is the first point that filtering is possible in DB2.

2. Index Screening = The Stage 1 predicate is a candidate for filtering on the index leaf pages. This is the second point of filtering in DB2.

3. Data Screening = The Stage 1 predicate is a candidate for filtering on the data pages. This is the third point of filtering in DB2.

4. Stage 2 = The predicate is not listed as Stage 1 and will be applied on the remaining qualifying pages from Stage 1. This is the fourth and final point of filtering in DB2.



Stage 1

value BETWEEN COL1 AND COL2 N N COL BETWEEN COL1 AND COL2 N N COL BETWEEN expression1 AND expression2 N N value NOT BETWEEN COL1 AND COL2 N N T1.COL1 op T1.COL2 N N T1.COL1 <> T1.COL2 N N COL = ANY (non subq) N N COL = ALL (non subq) N N COL <> ANY (non subq) N N COL <> ALL (non subq) N N COL NOT IN (non subq) N N COL = (cor subq) N N

Stage 2 Predicates - z/OS(Chapter 29, Table 123, page 734 of the Database Administration Guide)

V7 may promote

join Between



Stage 1

COL = (cor subq) N N COL = ANY (cor subq) N N COL = ALL (cor subq) N N COL op (cor subq) N N COL op ANY (cor subq) N N COL op ALL (cor subq) N N COL <> (cor subq) N N COL <> ANY (cor subq) N N COL <> ALL (cor subq) N N COL IN (cor subq) N N COL NOT IN (cor subq) N N EXISTS (subq) N N NOT EXISTS (subq) N N expression = value N N expression <> value N N expression op value N N expression op (subquery) N N

Stage 2 Predicates - z/OS(Chapter 29, Table 123, page 734 of the Database Administration Guide)

Not made Indexable,

.

.

V7 just rewrites

the query to a join

Four Points of Filtering – DB2Four Points of Filtering – DB21. Indexable Stage 1 Probe2. Stage 1 Index Filtering 3. Stage 1 Data Filtering4. Stage 2

Type 2 Index

Leaf Pageo o o o o o



Non-Leaf PageO O O



Non-Leaf PageO O





Non-Leaf PageO O O O

Root PageO O OC1.C2.C3

44

11

22

33

TOKEN_NR. ROLE_CD

WHERE C.LAST_NM LIKE ? AND C.TOKEN_NR =

B.TOKEN_NRC.ROLE_CD > ?

AND CASE C.SEX WHEN ‘X’ THEN ? END) = ‘ABCDE’

WHERE C.LAST_NM LIKE ? AND C.TOKEN_NR =

B.TOKEN_NRC.ROLE_CD > ?

AND CASE C.SEX WHEN ‘X’ THEN ? END) = ‘ABCDE’


SQL Performance Rules ….. Promote Stage 2’s/Residuals if possiblePromote Stage 2’s/Residuals if possible

» Value BETWEEN COL1 AND COL2 is Stage 2Value BETWEEN COL1 AND COL2 is Stage 2» Rewrite to: (Value >= COL1 AND value <= COL2)Rewrite to: (Value >= COL1 AND value <= COL2)

» COL NOT IN (K, S, T) = COL IN (known values)COL NOT IN (K, S, T) = COL IN (known values) SELECT only the columns neededSELECT only the columns needed

» Disallow SELECT *Disallow SELECT * SELECT only the rows neededSELECT only the rows needed

» Disallow program filteringDisallow program filtering Use constants and literals if the values will not Use constants and literals if the values will not

change in the next 3 yearschange in the next 3 years» Increase optimizer accuracy for static valuesIncrease optimizer accuracy for static values

WHERE ‘0001-01-01’ BETWEEN B.SCND_ALOC_EFF_DATE AND B.SCND_ALOC_END_DATE

Should be:

WHERE ‘0001-01-01’ >= B.SCND_ALOC_EFF_DATE AND ‘0001-01-01’ <= B.SCND_ALOC_END_DATE

Bad BETWEENs


…… SQL Performance Rules Make WHERE clause data types matchMake WHERE clause data types match

» Costs more to navigate indexes when not matchingCosts more to navigate indexes when not matching Do not SELECT columns with known static valuesDo not SELECT columns with known static values

» Disallow COLn from SELECT list if COLn = valueDisallow COLn from SELECT list if COLn = value Do not place any local filtering in the ON clause for OUTER Do not place any local filtering in the ON clause for OUTER

JOINs JOINs » Disallow “during join” filtering, encourage “before join”Disallow “during join” filtering, encourage “before join”

Sequence filtering from most restrictive to least restrictive by Sequence filtering from most restrictive to least restrictive by table, by predicate type table, by predicate type

WHEREWHERE A.COL2 = ‘abracadabra’A.COL2 = ‘abracadabra’ ANDAND A.COL1 IN (:hv1, :hv2, :hv3)A.COL1 IN (:hv1, :hv2, :hv3)ANDAND A.COL4 > :hvcol4 A.COL4 > :hvcol4 AND AND A.COL3 > :hvcol3A.COL3 > :hvcol3ANDAND A.COL5 LIKE ‘%SON’A.COL5 LIKE ‘%SON’

Ties

Matter!


Data Type ConversionsWHERE DATE(col_TS) BETWEEN :date1WHERE DATE(col_TS) BETWEEN :date1

AND :date2AND :date2

WHERE col_TS BETWEEN WHERE col_TS BETWEEN

TIMESTAMP(:date1, ’00:00:00’)TIMESTAMP(:date1, ’00:00:00’)

AND TIMESTAMP(:date2, ’59:59:99’)AND TIMESTAMP(:date2, ’59:59:99’)

Need to leave columns alone and translate the Need to leave columns alone and translate the second argumentsecond argument


Host Variable Filter Factors (DB2 Database Administration Guide- Properties of Predicates)

COLCARDF Factor for <, <=, >, >= Factor for LIKE or BETWEEN >=100,000,000 1/10,000 3/100,000 >=10,000,000 1/3,000 1/10,000 >=1,000,000 1/1,000 3/10,000 >=100,000 1/300 1/1,000 >=10,000 1/100 3/1,000 >=1,000 1/30 1/100 >=100 1/10 3/100 >=2 1/3 1/10 =1 1/1 1/1 <=0 1/3 1/10

COLCARD Factor for <, <=, >, >= BETWEEN, LIKE

>=100,000,000 1/100,000,000

>=10,000,000 1/10,000,000

>=1,000,000 1/1,000,000

>=100,000 1/100,000

>=10,000 1/10,000

>=1,000 1/1,000

>=100 1/100

>=0 1/3


SQL Review Checklist1.1. Examine Program logicExamine Program logic

2.2. Examine FROM clauseExamine FROM clause

3.3. Verify Join conditionsVerify Join conditions

4.4. Promote Stage 2’s/Residuals Promote Stage 2’s/Residuals

5.5. Prune SELECT listsPrune SELECT lists

6.6. Verify local filtering sequenceVerify local filtering sequence

7.7. Analyze Access PathsAnalyze Access Paths

8.8. Tune if necessaryTune if necessary

SQL Portfolio- DB2 9 z/OS vs. DB2 9.5 LUW

Inner and Outer Joins, Table Expressions, Subqueries, GROUP BY, Complex Inner and Outer Joins, Table Expressions, Subqueries, GROUP BY, Complex Correlation, Global Temporary Tables, CASE, 100+ Built-in Functions including Correlation, Global Temporary Tables, CASE, 100+ Built-in Functions including

SQL/XML, Limited Fetch, Insensitive Scrollable Cursors, UNION Everywhere, SQL/XML, Limited Fetch, Insensitive Scrollable Cursors, UNION Everywhere, MIN/MAX Single Index Support, Self Referencing Updates with Subqueries, Sort MIN/MAX Single Index Support, Self Referencing Updates with Subqueries, Sort Avoidance for ORDER BY, and Row Expressions 2M Statement Length, GROUP Avoidance for ORDER BY, and Row Expressions 2M Statement Length, GROUP

BY Expression, Sequences, Scalar Fullselect, Materialized Query Tables, BY Expression, Sequences, Scalar Fullselect, Materialized Query Tables, Common Table Expressions, Recursive SQL, CURRENT PACKAGE PATH, Common Table Expressions, Recursive SQL, CURRENT PACKAGE PATH,

VOLATILE Table Support, Star Join Sparse Index, Qualified Column names, VOLATILE Table Support, Star Join Sparse Index, Qualified Column names, Multiple DISTINCT clauses, ON COMMIT DROP, Transparent ROWID Column, Multiple DISTINCT clauses, ON COMMIT DROP, Transparent ROWID Column,

call from trigger, statement isolation, FOR READ ONLY KEEP UPDATE LOCKS, call from trigger, statement isolation, FOR READ ONLY KEEP UPDATE LOCKS, SET CURRENT SCHEMA, client special registers, long SQL Object names, SET CURRENT SCHEMA, client special registers, long SQL Object names,

SELECT FROM SELECT FROM INSERTINSERT, UPDATE, DELETE, MERGE, INSTEAD OF TRIGGER, , UPDATE, DELETE, MERGE, INSTEAD OF TRIGGER, Native SQL Procedure Language, BIGINT, file reference variables, XML, FETCH Native SQL Procedure Language, BIGINT, file reference variables, XML, FETCH

FIRST & ORDER BY IN subselect and fullselect, caseless comparisons, FIRST & ORDER BY IN subselect and fullselect, caseless comparisons, INTERSECT, EXCEPT, not logged tables, DECIMAL FLOAT, XQuery,TRUNCATE, INTERSECT, EXCEPT, not logged tables, DECIMAL FLOAT, XQuery,TRUNCATE,

OLAP Functions, Session variables, OmniFind, Spatial, ROLEOLAP Functions, Session variables, OmniFind, Spatial, ROLE


GROUPING SETS, ROLLUP, CUBE, Many Built-in Functions, SET CURRENT ISOLATION , multi-site join, MERGE, ARRAY data type,

global variables, Oracle syntax, XML enhancements

Stage1 unlike data types, Multi-row INSERT, FETCH, Multi-row cursor UPDATE, Dynamic Scrollable Cursors, Multiple CCSIDs per statement,

GET DIAGNOSTICS, Enhanced UNICODE, IS NOT DISTINCT FROM, VARBINARY, FETCH CONTINUE, MERGE

SQL Performance Rule Violations & Misuse Cost CPU & I/OCost CPU & I/O Cost time to fixCost time to fix Cost $$$$$$$$$Cost $$$$$$$$$

How much does it cost to misuse SQL?

1. Use joins over subqueries when detail row information is required2. Use subqueries over joins when detail row information is not required3. Use INNER JOIN over LEFT JOIN when exceptions are not expected or

needed4. Use CREATE GLOBAL TEMPORARY TABLE when data is infrequently

accessed5. Use DECLARE GLOBAL TEMPORARY TABLE with a clustered index

when DTT is large and data is frequently accessed

How much does it cost to misuse SQL?

1. Use joins over subqueries when detail row information is required2. Use subqueries over joins when detail row information is not required3. Use INNER JOIN over LEFT JOIN when exceptions are not expected or

needed4. Use CREATE GLOBAL TEMPORARY TABLE when data is infrequently

accessed5. Use DECLARE GLOBAL TEMPORARY TABLE with a clustered index

when DTT is large and data is frequently accessed

DB2 10 z/OS vs. DB2 10 LUW

Inner and Outer Joins, Table Expressions, Subqueries, GROUP BY, Complex Inner and Outer Joins, Table Expressions, Subqueries, GROUP BY, Complex Correlation, Global Temporary Tables, CASE, 100+ Built-in Functions including Correlation, Global Temporary Tables, CASE, 100+ Built-in Functions including

SQL/XML, Limited Fetch, Insensitive Scroll Cursors, UNION Everywhere, MIN/MAX SQL/XML, Limited Fetch, Insensitive Scroll Cursors, UNION Everywhere, MIN/MAX Single Index, Self Referencing Updates with Subqueries, Sort Avoidance for ORDER Single Index, Self Referencing Updates with Subqueries, Sort Avoidance for ORDER

BY, and Row Expressions, 2M Statement Length, GROUP BY Expression, BY, and Row Expressions, 2M Statement Length, GROUP BY Expression, Sequences, Scalar Fullselect, Materialized Query Tables, Common Table Sequences, Scalar Fullselect, Materialized Query Tables, Common Table

Expressions, Recursive SQL, CURRENT PACKAGE PATH, VOLATILE Tables, Star Expressions, Recursive SQL, CURRENT PACKAGE PATH, VOLATILE Tables, Star Join Sparse Index, Qualified Column names, Multiple DISTINCT clauses, ON Join Sparse Index, Qualified Column names, Multiple DISTINCT clauses, ON

COMMIT DROP, Transparent ROWID Column, Call from trigger, statement isolation, COMMIT DROP, Transparent ROWID Column, Call from trigger, statement isolation, FOR READ ONLY KEEP UPDATE LOCKS, SET CURRENT SCHEMA, Client special FOR READ ONLY KEEP UPDATE LOCKS, SET CURRENT SCHEMA, Client special

registers, long SQL object names, SELECT from INSERT, UPDATE or DELETE, registers, long SQL object names, SELECT from INSERT, UPDATE or DELETE, INSTEAD OF TRIGGER, Native SQL Procedure Language, BIGINT, file reference INSTEAD OF TRIGGER, Native SQL Procedure Language, BIGINT, file reference variables, XML, FETCH FIRST & ORDER BY in subselect & fullselect, caseless variables, XML, FETCH FIRST & ORDER BY in subselect & fullselect, caseless

comparisons, INTERSECT, EXCEPT, not logged tables, OmniFind, spatial, range comparisons, INTERSECT, EXCEPT, not logged tables, OmniFind, spatial, range partitions, data compression, session variables, DECIMAL FLOAT, optimistic partitions, data compression, session variables, DECIMAL FLOAT, optimistic

locking, ROLE, TRUNCATE, index & XML compression, created temps, inline LOB, locking, ROLE, TRUNCATE, index & XML compression, created temps, inline LOB, administrative privileges, administrative privileges, implicit cast, date/time changes, currently committed, implicit cast, date/time changes, currently committed, moving sum & average, index include columns, row and column access control, moving sum & average, index include columns, row and column access control,

time travel querytime travel query

21

Updateable UNION in Views, GROUPING SETS, ROLLUP, CUBE, more Built-in Functions, SET CURRENT ISOLATION, multi-site join, MERGE, MDC, XQuery, XML enhancements, additional data type (array, row, cursor), global variables, even more vendor syntax, temp table compression, MODULEs

Multi-row INSERT, FETCH & multi-row cursor UPDATE, Dynamic Scrollable Cursors, GET DIAGNOSTICS, Enhanced UNICODE SQL, join across encoding schemes, IS NOT DISTINCT FROM, VARBINARY, FETCH CONTINUE, MERGE, SELECT from MERGE, routine versioning, timestamps w/timezone

DB2 11 SQL – Standard SQL support (not exhaustive, some features may be missing)

DB2 11 for z/OS and DB2 10.5 Linux, Unix & Windows

Multi-row INSERT, FETCH & multi-row cursor UPDATE, Dynamic Scrollable Cursors, GET DIAGNOSTICS, Enhanced UNICODE SQL, join across encoding schemes, IS NOT DISTINCT FROM, VARBINARY, FETCH CONTINUE, SELECT FROM MERGE, MERGE, routine versioning, transparent archive query

Inner and Outer Joins, Table Expressions, Subqueries, GROUP BY, Complex Correlation, Global Temporary Tables, CASE, 100+ Built-in Functions including SQL/XML, Limited Fetch, Insensitive Scroll Cursors, UNION Everywhere, MIN/MAX Single Index, Self Referencing Updates with Subqueries, Sort Avoidance for ORDER BY, and Row Expressions, 2M Statement Length, GROUP BY Expression, Sequences, Scalar Fullselect, Materialized Query Tables, Common Table Expressions, Recursive SQL, CURRENT PACKAGE PATH, VOLATILE Tables, Star Join Sparse Index, Qualified Column names, Multiple DISTINCT clauses, ON COMMIT DROP, Transparent ROWID Column, Call from trigger, statement isolation, FOR READ ONLY KEEP UPDATE LOCKS, SET CURRENT SCHEMA, Client special registers, long SQL object names, SELECT from INSERT, UPDATE or DELETE, INSTEAD OF TRIGGER, SQL PL in routines, BIGINT, file reference variables, XML, FETCH FIRST & ORDER BY in subselect & fullselect, caseless comparisons, INTERSECT, EXCEPT, MERGE not logged tables, OmniFind, spatial, range partitions, data compression, DECFLOAT, optimistic locking, ROLE, TRUNCATE, index & XML compression, created temps, inline LOB, administrative privileges, implicit cast, increased timestamp precision, currently committed, moving sum & average, index include columns, row and column access controls, time travel query, GROUPING SETS, ROLLUP, CUBE, global variables, Text Search functions, accelerated tables, DROP COLUMN, array data type, XML enhancements

Updateable UNION in Views, more Built-in Functions, SET CURRENT ISOLATION, multi-site join, full MERGE, MDC, XQuery, additional data type (row, cursor), even more vendor syntax, temp table compression, MODULEs

z

luw

common

SQL Skill Self Assessment Level Assessment = YOU CAN FULLY UNDERSTAND THE FEATURE AND PROPER USE OF:

0 You think SQL is a new energy drink

1 Simple SELECT statements, WITH clause, ORDER BY

2

WHERE clauses, BETWEEN, LIKE, IN(list), =, >=, >, <, <=, <>, NOT IN(list), NOT LIKE, NOT BETWEEN, DISTINCT

3

Table joins (inner, outer, full), UNION, UNION ALL, CONCAT, static CURSORs, FOR UPDATE OF, ROW_NUMBER, EXCEPT, INTERCEPT

4

noncorrelated and correlated subqueries, EXISTS, NOT EXISTS, FETCH FIRST x ROWS ONLY, OPTIMIZE FOR x ROWS, MERGE, TRUNCATE

5

Indexable, Stage1 and Stage 2 predicate evaluation, multirow FETCH/INSERT, GET DIAGNOSTICS, Scalar full SELECT

6

Table expressions/common table expressions, GROUP BY, HAVING, IS NOT DISTINCT FROM, embedded dynamic SQL, ORDER BY/FETCH FIRST in subselect, FETCH CONTINUE

7

CASE expressions, Global Temporary Table (GTT), Declared Temporary Table (DTT), Dynamic Scrollable cursors, SEQUENCES/IDENTITY

8

Queries involving > 10 tables, INSERT within (SELECT , UPDATE, DELETE, MERGE), Star Schema, GROUP BY expression

9

MQT (Materialized Query Tables), Recursive SQL, UNION in Views, > 30 useful Built-in Functions, DENSE_RANK, RANK

10

Codes effective and efficient SQL applying performance rules and knows when to use each SQL feature appropriately

Name: __________________

Before:_______

After:________


When to Tune Queries Not until the query is coded the best it can beNot until the query is coded the best it can be All predicates are the best they can beAll predicates are the best they can be

» Promote Stage 2’s if possiblePromote Stage 2’s if possible» Promote Stage 1’s if possiblePromote Stage 1’s if possible» Apply performance rulesApply performance rules

Check Access Paths of all Query BlocksCheck Access Paths of all Query Blocks Apply data knowledge and program knowledge to Apply data knowledge and program knowledge to

predict response timepredict response time If, and only if, the predicted service levels are not If, and only if, the predicted service levels are not

met - TUNE!met - TUNE!


Why Tune Queries Performance!Performance! Host variables force the optimizer to guessHost variables force the optimizer to guess IBM optimizer researchers and developers IBM optimizer researchers and developers

provide extremely good guessesprovide extremely good guesses Occasionally (2-7% of queries) they guess Occasionally (2-7% of queries) they guess

wrongwrong Have no fear, changing the optimizer’s mind Have no fear, changing the optimizer’s mind

is easier than you think is easier than you think


How to Tune Queries Do not change statistics, just keep accurateDo not change statistics, just keep accurate Do not panicDo not panic Choose a proven, low maintenance, tuning Choose a proven, low maintenance, tuning

techniquetechnique IBM’s list:IBM’s list:

» OPTIMIZE FOR n ROWSOPTIMIZE FOR n ROWS» FETCH FIRST n ROWS ONLYFETCH FIRST n ROWS ONLY» No Op (+0, CONCAT ‘ ‘)No Op (+0, CONCAT ‘ ‘)» TX.CX=TX.CXTX.CX=TX.CX» REOPT(VARS) REOPT(VARS) » ON 1=1ON 1=1

SQL Tuning ExamplesWHERE S.SALES_ID > 44 AND S.MNGR = :hv-mngr AND S.REGION BETWEEN

:hvlo AND :hvhi CONCAT ‘ ‘


SELECT S.QTY_SOLD, S.ITEM_NO, S.ITEM_NAME

FROM SALE SWHERE S.ITEM_NO > :hvORDER BY ITEM_NOFETCH FIRST 22 ROWS ONLY

WHERE B.BID BETWEEN :hvlo AND :hvhi

AND B.BID = D.DID AND B.SID = S.SID AND B.COL2 >= :hv

AND B.COL3 >= :hv

AND B.COL4 >= :hv

No Operation

Limited Fetch

Fake Filter


Tuning Tools Sheryl’s Extended ListSheryl’s Extended List

» Fake FilteringFake Filtering• COL BETWEEN :hv1 AND :hv2COL BETWEEN :hv1 AND :hv2• COL >= :hvCOL >= :hv

» Table expressions with DISTINCTTable expressions with DISTINCT• FROM (SELECT DISTINCT COL1, COL2 …..)FROM (SELECT DISTINCT COL1, COL2 …..)

» Anti-JoinsAnti-Joins» Extreme ExperimentsExtreme Experiments» Index ChangesIndex Changes» MQT DesignMQT Design


What Used to Work – z/OS

UNION queries for index access – until V3 UNION queries for index access – until V3 Multiple Index AccessMultiple Index Access

(Predicate OR 0 = 1) – until V5 No Operations(Predicate OR 0 = 1) – until V5 No Operations Layered Table Expressions – until V6 smarter Layered Table Expressions – until V6 smarter

optimizeroptimizer Correlated Subqueries – until V7 query re-writeCorrelated Subqueries – until V7 query re-write No cross table indexes - until V8 Materialized No cross table indexes - until V8 Materialized

Query Tables (more on that later)Query Tables (more on that later)

30

Steps in Tuning Queries

Step 1 - Learn Access PathsStep 1 - Learn Access Paths

Step 2 – Interpret the Optimizer DecisionsStep 2 – Interpret the Optimizer Decisions

Step 3 – Find Top CPU ConsumersStep 3 – Find Top CPU Consumers

Step 4 – Learn Tuning TechniquesStep 4 – Learn Tuning Techniques

Step 5 – Tune if NecessaryStep 5 – Tune if Necessary

Step 6 – Designing Performance Step 6 – Designing Performance StructuresStructures

31

Step 1

Learn the DB2 Access PathsLearn the DB2 Access Pathswww.smlsql.com

www.ca.com/ Google Sheryl Larsen Google Sheryl Larsen


All the Possible Access Paths

Direct Row

Star

Merge Scan

Hybrid

Nested Loop

Partitioned Table Scan

Table Scan

Limited Partition ScanWith NPI

Limited Partition ScanWith Partitioning Index

Partition ScanWithout Index

Multiple Index Access

List Prefetch

NonMatchingIndex Access

Matching Index Access

One Fetch

IN(list) Index Access

Merge ScanMerge Scan

Star Join: Star Join: Cartesian Cartesian or or Pair-wisePair-wise

Hybrid Join: Hybrid Join: Type C or Type C or Type NType N

Nested LoopNested Loop

Limited Partition ScanLimited Partition ScanUsing Data Partitioned Using Data Partitioned Secondary IndexSecondary Index

Limited Partition ScanLimited Partition ScanUsing Partitioning IndexUsing Partitioning Index

Multiple Index AccessMultiple Index Access

List PrefetchList Prefetch

NonMatchingNonMatchingIndex AccessIndex Access

Matching Index AccessMatching Index Access

Sparse Index AccessSparse Index Access

One FetchOne Fetch

IN(list) Index Access IN(list) Index Access

(Bold names use an Index)


Variations of Index Accesses

One Fetch IN(list) Index Access


NonMatching Index AccessSparse Index Access

List Prefetch


Limited Partition Scan With Partitioning Index

Limited Partition ScanWith NPSI

Limited Partition ScanWith DPSI

Multidimensional Index Access

All can be parallelAll can be parallel All can omit data accessAll can omit data access Clustering index or Clustering index or

nonclustering accessnonclustering access




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Rid + value(s)

rang

e

Bid


One Fetch

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

SELECT MIN(C1)FROM T1

SELECT MIN(C1)FROM T1

SELECT MAX(C3)FROM T1WHERE C1 = ‘L’ AND C2 = 99

SELECT MAX(C3)FROM T1WHERE C1 = ‘L’ AND C2 = 99

C1.C2.C3

3311

22



Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

WHERE C1 = ‘L’ AND C2 > 99 AND C3 = :hv

WHERE C1 = ‘L’ AND C2 > 99 AND C3 = :hv

C1.C2.C311

22

33


NonMatching Index Access

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

WHERE C2 > 99 AND C3 = :hv


C1.C2.C3

11

22


Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O


WHERE C1 IN (‘K’, ‘S’, ‘T’) AND C2 > 99 AND C3 = :hv


C1.C2.C3

3311 22


Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

IN(list) Index Access -Parallel



C1.C2.C3

1111 11


List Prefetch



COL1 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

33

11

22

44



COL1 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

CO L2 Index



Non-Leaf PageO O



Non-Leaf PageO O



Non-Leaf PageO O

Root PageO O O

AND/OR

WHERE C1 > 50WHERE C1 > 50WHERE C2 = :hvWHERE C2 = :hv

C1 C2

33

11

2244

55

6677

88


Sparse Index Access

Use Any Single Table Access

MethodAnd Filter

Use Any Single Table Access

MethodAnd Filter

33

11 22


Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

C1 PartsPartitioning Index – Partitioning Index – ClusteredClustered

Limited Partition ScanUsing Clustered Partitioning Index



1111 11

2222 22

3333 33


WHERE CORP IN (20, 40,60) AND

TICKET_PREFIX = ‘A’ AND CUST_NAME

LIKE ‘%CAMB%’

WHERE CORP IN (20, 40,60) AND

TICKET_PREFIX = ‘A’ AND CUST_NAME

LIKE ‘%CAMB%’

Limited Partition ScanUsing Nonclustered Partitioning Index

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

CORPPartitioning Index – Partitioning Index – NonclusteredNonclustered

1111 11

2222 22

3333 33


WHERE TOKEN_NR BETWEEN ? AND ? AND TICKET_PREFIX = ‘A’ AND LAST_NM LIKE ‘LAR%’

WHERE TOKEN_NR BETWEEN ? AND ? AND TICKET_PREFIX = ‘A’ AND LAST_NM LIKE ‘LAR%’

Scan for Last NameUsing Nonclustered Partitioning Index

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

TOKEN_NR

Partitioning Index – Partitioning Index – NonclusteredNonclustered

1111 11

2222 22

3333 33


Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

TOKEN_NR PartsPartitioning Index – Partitioning Index –

ClusteredClustered

Limited Partition ScanUsing Clustered Partitioning Index

WHERE TOKEN_NR BETWEEN ? AND ? AND LAST_NM > 99 AND ROLE_CD = ?

WHERE TOKEN_NR BETWEEN ? AND ? AND LAST_NM > 99 AND ROLE_CD = ?

1111 11

2222 22

3333 33

11

33


Limited Partition Scan Using DPSI

DPSI = DPSI = DData ata PPartitioned artitioned SSecondary econdary IIndexndexClustered byClustered by

LAST_NMLAST_NM

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

TOKEN_NR

Partitioning by Partitioning by TOKEN_NRTOKEN_NR

111111

2222 22

3333 33

WHERE TOKEN_NR BETWEEN ? AND ?

AND LAST_NM LIKE ‘%LAR

WHERE TOKEN_NR BETWEEN ? AND ?

AND LAST_NM LIKE ‘%LAR


Helps Improve Performance

Partition/Cluster by Date

Partition by DateCluster by Custid

TOKEN_NR BETWEEN ? AND ?AND LAST_NM_LIKE ?old

New

Partition/Cluster by DatePartition/Cluster

by DatePartition/Cluster by DatePartition/Cluster

by DatePartition/Cluster by TOKEN_NR

Partition by DateCluster by CustidPartition by Date

Cluster by CustidPartition by DateCluster by CustidPartition by TOKEN_NRCluster by LAST_NM

BP03

TOKEN_NR BETWEEN ? AND ?AND LAST_NM_LIKE ?


Limited Partition Scan Using DPSI

Data PartitionedData PartitionedSecondary IndexSecondary Index

DPSIDPSI

Clustered byClustered byCustomerCustomer

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

C1 Parts

Partitioning by DatePartitioning by Date

111111

2222 22

3333 33


Variations of Table Access

All can be parallel All can be parallel

If not enough room in If not enough room in memory, at run time memory, at run time create sparse index create sparse index insteadinstead

SegmentedPartitioned

Limited PartitionedIn Memory Data Cache


Table Scan

11

WHERE C1 BETWEEN :lowest AND

:highest


:highest



Part 1Part 1 Part 2 Part 3 Part 4 Part 5 Part 6 Part 7 Part 8

11


:highest


:highest


Limited Partitioned Table Scan

Part 1Part 1 Part 3 Part 4 Part 16 Part 17 Part 18

11

WHERE C1 IN (1, 3, 4, 16, 17, 18)WHERE C1 IN (1, 3, 4, 16, 17, 18)


Variations of Join Methods

All choose outer table and All choose outer table and filter firstfilter first

All can be parallel (Star All can be parallel (Star CPU only )CPU only )

Worry about join table Worry about join table sequence instead of join sequence instead of join methodmethod

Nested LoopHybrid Join Type CHybrid Join Type NMerge Scan Join

Star Join – CartesianStar Join – Pair Wise


COL1 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Inner Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

11

22

33

COL1 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Nested Loop Join

44

WHERE T1.ORDER >

222 AND

T1.ID = T2.IDAND

T2.COLOR IN (‘BLUE’,

‘YELLOW’, ‘GREEN’)

WHERE T1.ORDER >

222 AND

T1.ID = T2.IDAND

T2.COLOR IN (‘BLUE’,

‘YELLOW’, ‘GREEN’)

ID.COLOR index


Filtered outerRIDS

RIDS

Hybrid Join – Type N

22

CO L2 Index



Non-Leaf PageO O



Non-Leaf PageO O



Non-Leaf PageO O

Root PageO O O

44

11

33

COL1 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

33

44

55

66


11

22

33

CO L2 Index



Non-Leaf PageO O



Non-Leaf PageO O



Non-Leaf PageO O

Root PageO O O

Hybrid Join – Type CCOL1 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

44


Merge Scan Join

2211

33

44

WHERE T1.C1 = T2.CA ANDT1.C2 = T2.CB AND

T1.C3 = T2.CC

WHERE T1.C1 = T2.CA ANDT1.C2 = T2.CB AND

T1.C3 = T2.CC


Star Join - Cartesian

33

66

11

44

55

Inner Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

22

Big/Fact Table Index

Small/Dimension Tables


Pair-Wise Star Join

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O 33

66

11

44

55

22

77

99

88


Read Direct Access 1.1. Create table with ROWID Create table with ROWID

type column (DIRECT_ID)type column (DIRECT_ID)

2.2. SELECT DIRECT_IDSELECT DIRECT_IDINTO INTO :direct-id:direct-idFROM TAB12FROM TAB12WHERE UKEY = ‘AMX’WHERE UKEY = ‘AMX’

3.3. UPDATE TAB12UPDATE TAB12WHERE WHERE DIRECT_ID = DIRECT_ID = :direct-id:direct-id

DIRECT_ID

AMX DIRECT_ID

DIRECT_ID

DIRECT_ID33

11

22


All the Possible Access Paths

Direct Row

Star

Merge Scan

Hybrid

Nested Loop


Table Scan

Limited Partition ScanWith NPI

Limited Partition ScanWith Partitioning Index

Partition ScanWithout Index


List Prefetch

NonMatchingIndex Access


One Fetch


Merge ScanMerge Scan

Star Join: Star Join: Cartesian Cartesian or or Pair-wisePair-wise

Hybrid Join: Hybrid Join: Type C or Type C or Type NType N

Nested LoopNested Loop

Limited Partition ScanLimited Partition ScanUsing Data Partitioned Using Data Partitioned Secondary IndexSecondary Index

Limited Partition ScanLimited Partition ScanUsing Partitioning IndexUsing Partitioning Index

Multiple Index AccessMultiple Index Access

List PrefetchList Prefetch

NonMatchingNonMatchingIndex AccessIndex Access

Matching Index AccessMatching Index Access

Sparse Index AccessSparse Index Access

One FetchOne Fetch

IN(list) Index Access IN(list) Index Access

62

Step 2

Interpret the Optimizer’s Interpret the Optimizer’s DecisionsDecisions


Determine If Tuning Is Necessary

Find out what the general access path is for each Find out what the general access path is for each query block (subselect, union, table expression)query block (subselect, union, table expression)

Calculate execution frequencies of each query Calculate execution frequencies of each query block block » Correlated subselects execute bottom query block once Correlated subselects execute bottom query block once

for each qualifying row in bottom queryfor each qualifying row in bottom query Identify join table sequenceIdentify join table sequence Identify materialization (intermediate or final)Identify materialization (intermediate or final) Identify sorts (even hidden Rid sorts an IN(list)Identify sorts (even hidden Rid sorts an IN(list) Identify User Defined Table or Column functionsIdentify User Defined Table or Column functions


Using EXPLAIN

QQ

NoNo

QblockQblock

NoNo

StepStep

NoNo

MethMeth AccessAccess

TypeType

MatchMatch

ColsCols

PrePre

FetchFetch

JoinJoin

TypeType

IXIX

OnlyOnly

IXIX

NameName

11 11 11 00 RR 00 SS NN

11 11 22 44 II 22 LL NN IX22IX22

22 11 11 00 NN 22 YY

22 11 22 22 II 22 FF NN IX27IX27

22 11 33 44 II 22 LL NN IX42IX42

33 11 11 00 II 44 NN IXF1IXF1

33 22 11 22 RR 00 SS NN

EXPLAIN PLAN SET QUERYNO = n FOR …..;SELECT * FROM PLAN_TABLE ORDER BY …..;


Using EXPLAIN

SORTNSORTN

U J O GU J O G

SORTCSORTC

U J O GU J O G

PagePage

RangeRange

ColCol

FuncFunc

CorrCorr

NameName

QblockQblock

TypeType

OptOpt

HintHint

HintHint

UsedUsed

TSTS RemarkRemarkss

N N N N N N N N N N Y NN N Y N YY SS ABAB SELECTSELECT GGGG GGGG …….. Good 1Good 1

N N N NN N N N N N N NN N N N SS CDCD UPDATEUPDATE NNNN …….. Bad 1 Bad 1

N Y N NN Y N N N N N NN N N N YY EFEF INSERTINSERT ……..

N N N NN N N N N N N NN N N N YY RR XXXX DELECTDELECT …….. +0+0

N N N NN N N N N N N NN N N N YY GAGA NCOSUBNCOSUB XXXX …….. Bogus 3 Bogus 3

N N N NN N N N N N N NN N N N YY RR SLSL CORSUBCORSUB XXXX XXXX …….. Extra > Extra >

N N N NN N N N N N N NN N N N MLML DELCURDELCUR ……..

N N N NN N N N N N N NN N N N RR KLKL UPDCURUPDCUR …….. parensparens

N N N NN N N N N N N NN N N N SS TLTL SELUPDSELUPD SSSS SSSS …….. helphelp


Learning EXPLAIN Need IBM’s definition of the PLAN_TABLENeed IBM’s definition of the PLAN_TABLE Need a way to transform PLAN_TABLE into Need a way to transform PLAN_TABLE into

something readable (i.e. QMF Form or something readable (i.e. QMF Form or program)program)

Review PLAN_TABLE contents with Access Review PLAN_TABLE contents with Access Path descriptionsPath descriptions

Study, practice, practice, practiceStudy, practice, practice, practice


Using DSN_STATEMENT_TABLE

A = No Defaults used in estimateA = No Defaults used in estimate B = Check Reason columnB = Check Reason column

» Host variblesHost varibles» CardinalityCardinality» UDFUDF» TriggersTriggers» RIRI

Explain Explain TimeTime

StmtStmt

TypeType

Cost Cost

CatCat

PROCMSPROCMS PROCSUPROCSU ReasonReason

TimeTime

StampStamp

SELSEL

UPD…UPD…

A orA or

BB

MilliMilli

secondsseconds

ServiceService

UnitsUnits

For Cat For Cat BB


Service Units http://www-03.ibm.com/servers/eserver/zseries/srm/

zSeries 990

Processor Model STIDP Type STSI Model Name#C

PSU/SEC SRMsec/RealSec

2084-301 2084 301 1 21857.9235 508.1298

2084-302 2084 302 2 20752.2698 508.1298

2084-303 2084 303 3 20075.2823 508.1298

2084-304 2084 304 4 19559.9022 508.1298


Service Units

2097-701 2097 701 1 47619.0476 1106.1941

2097-702 2097 702 2 44692.7374 1106.1941

2097-703 2097 703 3 43010.7527 1106.1941

2097-704 2097 704 4 41666.6667 1106.1941

2097-501 2097 501 1 24427.4809 568.1815

2097-502 2097 502 2 23021.5827 568.1815

2097-503 2097 503 3 22222.2222 568.1815

Processor Model STIDP TypeSTSI Model

Name#CP SU/SEC SRMsec/RealSec

2097-401 2097 401 1 11291.4608 262.6049

2097-402 2097 402 2 10680.9079 262.6049

IBM System z10

SQL Tuning using EXPLAIN EXPLAIN ALL Expensive Queries EXPLAIN ALL Expensive Queries

-- Query Joins PLAN_TABLE -- Query Joins PLAN_TABLE with DSN_STATEMENTwith DSN_STATEMENT

BLOCK PLAN METH TABLE_NAME PQB BLOCK PLAN METH TABLE_NAME PQB TYPETYPE MCMC ANAME IO ANAME IO

--+---------+---------+---------+---------+---------+---------+---------+---------+---------+---------+---------+---------+---------+---------+---------+-------

01 01 00 R10_CLM_FACT12 12 01 01 00 R10_CLM_FACT12 12 M M 00 N 00 N

01 01 00 R10_CLM_FACT12 12 01 01 00 R10_CLM_FACT12 12 MX MX 01 01 X011_10_12_F_IX07X011_10_12_F_IX07 Y Y

01 01 00 R10_CLM_FACT12 12 01 01 00 R10_CLM_FACT12 12 MXMX 01 X011_10_12_F_IX04 Y 01 X011_10_12_F_IX04 Y

01 01 00 R10_CLM_FACT12 12 01 01 00 R10_CLM_FACT12 12 MI MI 00 N 00 N

01 01 00 R10_CLM_FACT12 12 01 01 00 R10_CLM_FACT12 12 MX MX 01 X011_10_12_F_IX04 Y 01 X011_10_12_F_IX04 Y

01 01 00 R10_CLM_FACT12 12 MX 01 X011_10_12_F_IX04 01 01 00 R10_CLM_FACT12 12 MX 01 X011_10_12_F_IX04 YY

01 01 00 R10_CLM_FACT12 12 MU 00 N 01 01 00 R10_CLM_FACT12 12 MU 00 N

01 01 00 R10_CLM_FACT12 12 MI 00 N 01 01 00 R10_CLM_FACT12 12 MI 00 N

01 01 0202 01 OTR_PRTY_LIAB_NET 12 N 03 OPL_NET_IX04 N 01 OTR_PRTY_LIAB_NET 12 N 03 OPL_NET_IX04 N

01 03 01 MEMBER_ELIG_NET 12 N 04 MBR_ELIG_NET_IX06 Y 01 03 01 MEMBER_ELIG_NET 12 N 04 MBR_ELIG_NET_IX06 Y

01 04 01 R10_CHG_FACT12 12 N 03 X007_10_12_F_IX19 N 01 04 01 R10_CHG_FACT12 12 N 03 X007_10_12_F_IX19 N

01 05 01 05 0101 R10_MULTI_CD_FAC12 12 I 02 X021_10_12_F_IX14 N R10_MULTI_CD_FAC12 12 I 02 X021_10_12_F_IX14 N

02 01 00 R10_CHG_FACT12 09 N 03 X007_10_12_F_IX19 Y 02 01 00 R10_CHG_FACT12 09 N 03 X007_10_12_F_IX19 Y

[email protected]©Copyright Sheryl Larsen 2011

[email protected] 70

Type of Access

# of Matching Columns

Index Name

Index Only?

Step #

Join Method

Sort, Prefetch, & Partition Usage Columns Sort N = New/Inner SortSort N = New/Inner Sort Sort C = Outer/Composite SortSort C = Outer/Composite SortSNSNUU SN SNJJ SN SNOO SN SNGG SCU SCJ SCO SCG PF QB_TYPE STMT_TYPE T_TP SCU SCJ SCO SCG PF QB_TYPE STMT_TYPE T_TP PRPR JT JT

+---------+---------+---------+---------+---------+---------+---------+-------+---------+---------+---------+---------+---------+---------+---------+-------

N N N N N N N N L N N N N N N N N L NCOSUBNCOSUB SELECT T SELECT T

N N N N N N N N S NCOSUB SELECT T N N N N N N N N S NCOSUB SELECT T


N N N N N N N N NCOSUB SELECT T N N N N N N N N NCOSUB SELECT T


N N N N N N N N N N N N N N N N SS NCOSUB SELECT T NCOSUB SELECT T







N N N N N N N N CORSUB SELECT T N N N N N N N N CORSUB SELECT T

©Copyright Sheryl Larsen 2011

Type of Query Block

Limited Partition Scan?

Prefetch Type

Unique

Join

Order

By

Group

By

Parallelism and Cost Columns Sort N = New/Inner SortSort N = New/Inner Sort Sort C = Outer/Composite SortSort C = Outer/Composite SortPM ADGREE JDGREE COST_CAT PROCMS PM ADGREE JDGREE COST_CAT PROCMS PROCSU PROCSU

--+---------+---------+---------+---------+---------+----+---------+---------+---------+---------+---------+--

C C 4 4 ------ ------ B B 710 21495 710 21495

C ------ ------ B 710 21495 C ------ ------ B 710 21495

C ------ ------ B 710 21495 C ------ ------ B 710 21495

C C ------ ------ B 710 21495 ------ ------ B 710 21495

C ------ ------ B 710 21495 C ------ ------ B 710 21495

C ------ ------ B 710 21495 C ------ ------ B 710 21495

C ------ ------ B 710 21495 C ------ ------ B 710 21495

C ------ ------ B 710 21495 C ------ ------ B 710 21495

C 4 4 B 710 21495 C 4 4 B 710 21495

C 4 4 B 710 21495 C 4 4 B 710 21495

C 4 4 B 710 21495 C 4 4 B 710 21495

C 4 4 B 710 21495 C 4 4 B 710 21495



Processor Service Units

# of I/O Streams

Contains :hostvars or cardinality issue

CPU Parallelism

73

Step 3

Fetch Top CPU ConsumersFetch Top CPU Consumers

Tuning SQL FIND ALL Expensive Queries FIND ALL Expensive Queries

---------+---------+------------+---------+---

PROGNAME PROCSU PROGNAME PROCSU

---------+---------+------------+---------+---

EXPNPROG 121,059,664 EXPNPROG 121,059,664 EXPNPROG 21,059,664 EXPNPROG 21,059,664

ONESECPG 79,664 ONESECPG 79,664

SUBSECPG SUBSECPG 9,664 9,664

CHEEPPRG CHEEPPRG 64 64

FREEPROG FREEPROG 4 4



PROCSU isTo Expensive to Calculate!


75

2,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,6472,147,483,647

76

Step 4

Learn Traditional and Extreme Learn Traditional and Extreme Tuning TechniquesTuning Techniques


Tuning Techniques to Apply When Necessary

Experiment with Extreme Tuning TechniquesExperiment with Extreme Tuning TechniquesDISTINCT Table ExpressionsDISTINCT Table ExpressionsCorrelated Table ExpressionsCorrelated Table ExpressionsOdd/old TechniquesOdd/old TechniquesAnti-JoinsAnti-JoinsManual Query Rewrite (X2QBOpt) covered in Manual Query Rewrite (X2QBOpt) covered in

ExtremeExtreme

Learn Traditional Tuning TechniquesLearn Traditional Tuning TechniquesOPTIMIZE FOR n ROWSOPTIMIZE FOR n ROWSNo OpsNo OpsREOPT(ONCE), REOPT(ALWAYS)REOPT(ONCE), REOPT(ALWAYS)Fake FilteringFake FilteringON 1 = 1ON 1 = 1Index & MQT DesignIndex & MQT Design


OPTIMIZE FOR n ROWSFETCH FIRST n ROWS

Both clauses influence the OptimizerBoth clauses influence the Optimizer» To encourage index access and nested loop join To encourage index access and nested loop join

» To discourage list prefetch, sequential prefetch, and To discourage list prefetch, sequential prefetch, and access paths with Rid processingaccess paths with Rid processing

» Use FETCH n = total rows required for setUse FETCH n = total rows required for set

» Use OPTIMIZE n = number of rows to send across Use OPTIMIZE n = number of rows to send across network for distributed applicationsnetwork for distributed applications

» Works at the statement levelWorks at the statement level


Fetch First Example Optimizer choose List Prefetch Index Optimizer choose List Prefetch Index

Access + sort for ORDER BY for 50,000 Access + sort for ORDER BY for 50,000 rowsrows

All qualifying rows processed All qualifying rows processed (materialized) before first row returned = (materialized) before first row returned = .81 sec.81 sec

<.1sec response time required<.1sec response time required


FROM SALE SWHERE S.ITEM_NO > :hvORDER BY ITEM_NO


FROM SALE SWHERE S.ITEM_NO > :hvORDER BY ITEM_NOFETCH FIRST 22 ROWS ONLY

• Optimizer now chooses Matching Index Access (first probe .004 sec)

• No materialization• Cursor closed after 22 items displayed (22 * .0008

repetitive access)• .004 + .017 = .021 sec


No Operation (No Op) +0, CONCAT ‘ ‘ also –0, *1, /1+0, CONCAT ‘ ‘ also –0, *1, /1

» Place no op next to predicatePlace no op next to predicate

» Use as many as neededUse as many as needed

» Discourages index access, however, preserves Discourages index access, however, preserves Stage 1Stage 1

» Can Alter table join sequenceCan Alter table join sequence

» Can fine tune a given access pathCan fine tune a given access path

» Can request a table scanCan request a table scan

» Works at the predicate levelWorks at the predicate levelDoes not Benefit

DB2 on Linux, UNIX or Windows


No Op Example CONCAT ‘ ‘

• Optimizer chooses Multiple Index Access• The table contains 100,000 rows and there are only 6 regions• Region range qualifies 2/3 of table• <.1sec response time required• No Op allows Multiple Index Access to continue on first 2 indexes• Two Matching index accesses, two small Rid sorts, & Rid intersection• ____unique Rids/16 * .000375 sec

= _____ + sort for Region = ______


FROM SALE SWHERE S.SALES_ID > 44 AND S.MNGR = :hv-mngr AND S.REGION BETWEEN

:hvlo AND :hvhiORDER BY S.REGION

SALES_ID.MNGR.REGION IndexSALES_ID.MNGR.REGION Index MNGR IndexMNGR Index REGION IndexREGION Index

…….FROM SALE SWHERE S.SALES_ID > 44 AND S.MNGR = :hv-mngr AND S.REGION BETWEEN

:hvlo AND :hvhi CONCAT ‘ ‘ORDER BY R.REGION


No Op Example - Scan

• If you know the predicates do very little filtering, force a table scan

• Use a No Op on every predicate• This forces a table scan• FOR FETCH ONLY encourages parallelism• WITH UR for read only tables to reduce CPU


FROM SALE SWHERE S.SALES_ID > 44 +0 AND S.MNGR = :hv-mngr CONCAT ‘ ‘ AND S.REGION BETWEEN

:hvlo AND :hvhi CONCAT ‘ ‘ORDER BY S.REGIONFOR FETCH ONLYWITH UR

SALES_ID.MNGR.REGION IndexSALES_ID.MNGR.REGION Index MNGR IndexMNGR Index REGION IndexREGION Index

Should this be Documented?

83

REOPT Package/Plan level optionPackage/Plan level option

QUERYOPT-- optimization-level QUERYOPT-- optimization-level • REOPT ONCEREOPT ONCE• REOPT ALWAYS REOPT ALWAYS • REOPT AUTO – DB2 9REOPT AUTO – DB2 9

– Every query’s access path gets re-optimized with host variable contents Every query’s access path gets re-optimized with host variable contents known only when neededknown only when needed

– System Parameter REOPTEXT System Parameter REOPTEXT enables REOPT(AUTO) for allenables REOPT(AUTO) for allunspecified dynamic queries unspecified dynamic queries with parameter markerswith parameter markers

» Long running queries/programs easily benefitLong running queries/programs easily benefit» High volume transactions will reflect the cost of the re-bind (90% of a full High volume transactions will reflect the cost of the re-bind (90% of a full

BIND)BIND)

DynamicStatement

Cache SELECT FROM T1 WHERE C1 > ? , 120

Compares New Values For BIG differences


Fake Filtering Fake PredicatesFake Predicates

» To encourage index accessTo encourage index access

» To alter table join sequence when nothing else To alter table join sequence when nothing else worksworks

» Works by decreasing filter factor on a certain tableWorks by decreasing filter factor on a certain table

» The filtering is fake and negligible costThe filtering is fake and negligible cost

» Not effective for dynamic queries if the filter contains Not effective for dynamic queries if the filter contains :host variables:host variables


Fake Filtering Example

SELECT B.BID, D.DID, S.SID,,D.DESC,, S.DESC

FROM BONDS B, DENOM D, SERIAL S


AND B.DID = D.DID AND B.SID = S.SIDORDER BY B.BID

B.BID Index6 billion rows


D.DID Index12 rows

D.DID Index12 rows

S.SID Index20 rows

S.SID Index20 rows

•Large report query with average of 400,000 row range of BID table•Need to start nested loop with big table•Tools required

Nested Loop

Joins


Fake Filtering Example

SELECT B.BID, D.DID, S.SID,,D.DESC,, S.DESC

FROM BONDS B, DENOM D, SERIAL S


AND B.BID = D.DID AND B.SID = S.SID

ORDER BY B.BID



D.DID Index12 rows

D.DID Index12 rows

S.SID Index20 rows

S.SID Index20 rows

•Keep adding filters until table join sequence changes•Start with index columns

•To preserve index-only access•No limit!

Nested Loop

Joins

AND B.COL6 >= :hv

AND B.COL2 >= :hv

AND B.COL3 >= :hv

AND B.COL4 >= :hv

AND B.COL5 >= :hv

= B.COL2

= B.COL3

= B.COL4

= B.COL5

= B.COL6

For Dynamic


ON 1 = 1 ON 1=1ON 1=1

» To fill in a required join fieldTo fill in a required join field

» To request a star join To request a star join

» When table ratios are under When table ratios are under the system specified number the system specified number (starts at 1:25)(starts at 1:25)

» Can benefit when large table Can benefit when large table has high selectivityhas high selectivity

88

Experiment with Extreme Techniques

After Traditional Techniques FailAfter Traditional Techniques Fail


DISTINCT Table Expressions Table expressions with DISTINCTTable expressions with DISTINCT

• FROM (SELECT DISTINCT COL1 FROM T1 …..) AS FROM (SELECT DISTINCT COL1 FROM T1 …..) AS STEP1STEP1 JOIN T2 ON … JOIN T3 ON …. JOIN T2 ON … JOIN T3 ON ….

» Used for forcing creation of logical set of dataUsed for forcing creation of logical set of data• No physical materialization if an index satisfies DISTINCTNo physical materialization if an index satisfies DISTINCT

» Can encourage sequential detectionCan encourage sequential detection

» Can encourage a Merge Scan joinCan encourage a Merge Scan join

Buffer Pool

Data

Index

WorkFile

STEP1 PhysicalSTEP1 Physical

STEP1 LogicalSTEP1 Logical


DISTINCT Table Expressions Example

SELECT ColumnsSELECT ColumnsFROM FROM TABXTABX, , TABYTABY,,

((SELECT DISTINCT COL1, COL2 …..SELECT DISTINCT COL1, COL2 ….. FROM FROM BIG_TABLE ZBIG_TABLE Z WHERE local conditions WHERE local conditions)) AS BIGZ AS BIGZ

WHERE join conditionsWHERE join conditions

Optimizer is forced to analyze the table expression prior to Optimizer is forced to analyze the table expression prior to joining TABX & TABYjoining TABX & TABY


Cardinality 1

Typical Join ProblemSELECT COL1, COL2 …..SELECT COL1, COL2 …..

FROM FROM ADDRADDR, , NAMENAME, , TAB3, TAB4, TAB5, TAB3, TAB4, TAB5, TAB6,TAB6, TAB7 TAB7 WHERE join conditionsWHERE join conditions

AND AND TAB6.CODE = :hvTAB6.CODE = :hv

Result is only 1,000 rowsResult is only 1,000 rows ADDR and NAME first two tables in joinADDR and NAME first two tables in join Index scan on Index scan on TAB6TAB6 table table

» Not good because zero filterNot good because zero filter


Keeps large tables joined last

Gets rid of Index Scan

Tuning TechniqueSELECT COL1, COL2 …..SELECT COL1, COL2 …..

FROM FROM ADDRADDR, , NAMENAME, ,

(SELECT DISTINCT columns(SELECT DISTINCT columns

FROM FROM TAB3, TAB4, TAB5, TAB3, TAB4, TAB5, TAB6,TAB6, TAB7 TAB7


AND (AND (TAB6.CODE = :hv OR 0 = 1)) TAB6.CODE = :hv OR 0 = 1)) AS TEMPAS TEMP



Correlated Table Expressions Advantage over correlated subquery Advantage over correlated subquery

» Access both sets of dataAccess both sets of data Performance gainsPerformance gains

» Access path changes Access path changes » Encourages Nested Loop JoinEncourages Nested Loop Join» Reduced processing and/or materializationReduced processing and/or materialization

SELECT Columns from either set of dataFROM table1 T1 ,TABLE(SELECT T2.STUDENT_YEAR, SUM(C8) AS SUM8, MAX(C9) AS MAX9

FROM table2 T2WHERE T1.column = T2.columnGROUP BY T2.STUDENT_YEAR) AS T2


Correlated Table Expressions

Get a student’s class load, and average & Get a student’s class load, and average & max class load of their peersmax class load of their peers

SELECT SR.NAME, SS.CLASS_LOAD, P.AVG_CLASS_LOAD, P.MAX_CLASS_LOADFROM SMLU_STUDENT_ROSTER SR

, TABLE(SELECT SR2.STUDENT_YEAR,AVG(SR2.NUM_OF_CLASSES) AS AVG_CLASS_LOAD,MAX(SR2.NUM_OF_CLASSES) AS MAX_CLASS_LOADFROM SMLU_STUDENT_ROSTER SR2WHERE SR.STUDENT_YEAR = SR2.STUDENT_YEARGROUP BY SR2.STUDENT_YEAR) AS P

LEFT JOIN TABLE(SELECT SS.SID, COUNT(*) AS SS.CLASS_LOADFROM SMLU_STUDENT_SCHED SSWHERE SS.SID = SR.SIDGROUP BY SS.SID) as SS

ON 1=1


Anti-Join

Inner Join

TABX Exceptions

TABY Exceptions

SELECT ColumnsSELECT Columns

FROM FROM TABX XTABX X

WHERE NOT EXISTSWHERE NOT EXISTS

(SELECT *(SELECT *

FROM FROM TABY YTABY Y

WHERE WHERE XX.COL1 = .COL1 = YY.COL1).COL1)



LEFT JOIN LEFT JOIN TABX XTABX X

ON ON XX.COL1 = .COL1 = YY.COL1.COL1

WHERE WHERE XX.COL1 IS NULL .COL1 IS NULL

SELECT ColumnsSELECT ColumnsFROM FROM TABX XTABX X, , TABY YTABY YWHERE WHERE XX.COL1 = .COL1 = YY.COL1.COL1


Stage 2 when correlated

Anti-JoinSELECT ColumnsSELECT Columns


WHERE NOT EXISTSWHERE NOT EXISTS

(SELECT *(SELECT *


WHERE WHERE XX.COL1 = .COL1 = YY.COL1).COL1)

Indexable Stage 1



LEFT JOIN LEFT JOIN TABY YTABY Y

ON ON XX.COL1 = .COL1 = YY.COL1.COL1

WHERE WHERE YY.COL1 IS NULL .COL1 IS NULL

TABX Exceptions

Does not Benefit LUW

97

Step 5

Tune if NecessaryTune if Necessary


Hunting Query IdeasWHERE WHERE PROGNAME LIKE ‘online%’PROGNAME LIKE ‘online%’

ANDAND ACCESSTYPE = ‘I’ ACCESSTYPE = ‘I’ ANDAND PREFETCH = ‘L’PREFETCH = ‘L’

WHERE WHERE ACCESSTYPE = ‘I’ ACCESSTYPE = ‘I’ ANDAND MATCH_COLS = 0MATCH_COLS = 0

WHEREWHERE METHOD = 4METHOD = 4ANDAND SORTN_JOIN = ‘Y’SORTN_JOIN = ‘Y’

WHEREWHERE ACCESSTYPE = ‘R’ACCESSTYPE = ‘R’

WHERE WHERE TNAME LIKE ‘DSNWF%’TNAME LIKE ‘DSNWF%’WHERE WHERE QBLOCK_TYPE = ‘TABLEX’QBLOCK_TYPE = ‘TABLEX’

ANDAND TABLE_TYPE = ‘W’TABLE_TYPE = ‘W’

List OnlinesDoing

List Prefetch

List IndexScans

List NastyHybrid Joins

List Table Scans

Materialization

Table Expression Materialization(V7)


SYSPACKSTMTWHERE WHERE STMT LIKE ‘%WHERE_:%BETWEEN%’STMT LIKE ‘%WHERE_:%BETWEEN%’

WHERE WHERE STMT LIKE ‘%WHERE%NOT%’STMT LIKE ‘%WHERE%NOT%’

WHEREWHERE STMT LIKE ‘%WHERE%COLX =%’STMT LIKE ‘%WHERE%COLX =%’

WHEREWHERE STMT LIKE ‘STMT LIKE ‘any SQL code problem any SQL code problem previously determinedpreviously determined’’

List UglyBETWEEN

clauses

List NOTs

Check dataTypes of literals

and constants

100

Step 6

Designing Optimal Designing Optimal Performance Structures for Performance Structures for One Query or a WorkloadOne Query or a Workload


One Query at a Time

Type 2 Index




Non-Leaf PageO O O



Non-Leaf PageO O






Root PageO O O

SELECT C7, C8, C12, C2, C3WHERE C4 = ‘L’ AND C5 = 99 AND C6 = :hv AND C1 IN (:a,:b,:c) AND C2 NOT LIKE :hv AND C3 BETWEEN

:hvstart AND :hvend AND C9 = (CASE expression)ORDER BY C1.C2.C3

SELECT C7, C8, C12, C2, C3WHERE C4 = ‘L’ AND C5 = 99 AND C6 = :hv AND C1 IN (:a,:b,:c) AND C2 NOT LIKE :hv AND C3 BETWEEN

:hvstart AND :hvend AND C9 = (CASE expression)ORDER BY C1.C2.C3

C4.C5.C6.C1.C2.C3.C9.C7.C8.C12

1.1. Equal predicates firstEqual predicates first2.2. Sequencing columns nextSequencing columns next3.3. Predicates most restrictive to Predicates most restrictive to

least restrictiveleast restrictive

4.4. All remaining SELECT ColumnsAll remaining SELECT Columns


…. To Build or not to BuildGeneric MQT strategiesGeneric MQT strategies» HUGE MQT with many indexesHUGE MQT with many indexes

» Many small MQTs with few indexesMany small MQTs with few indexes

Optimal MQT strategyOptimal MQT strategy» Most popular queries as MQTsMost popular queries as MQTs

• Common tablesCommon tables• Common predicate setsCommon predicate sets• Common data translationsCommon data translations

» Indexes on the popular ones to allow access to subset of Indexes on the popular ones to allow access to subset of MQTMQT


Expensive Recursive SyntaxWITH ORG_CHART(BOSS, DIRECT_REPORT, N) ASWITH ORG_CHART(BOSS, DIRECT_REPORT, N) AS

(SELECT (SELECT BOSS, DIRECT_REPORT, 1BOSS, DIRECT_REPORT, 1

FROM FROM ORG_TABLEORG_TABLE

WHEREWHERE BOSS= ‘SAM’BOSS= ‘SAM’

UNION ALLUNION ALL

SELECT SELECT T1.BOSS, T2.DIRECT_REPORT, N+1T1.BOSS, T2.DIRECT_REPORT, N+1

FROM FROM ORG_CHART T2, ORG_TABLE T1ORG_CHART T2, ORG_TABLE T1

WHEREWHERE T2.DIRECT_REPORT = T2.BOSST2.DIRECT_REPORT = T2.BOSS

ANDAND N < 10 0,000)N < 10 0,000)

SELECT BOSS, DIRECT_REPORT, NSELECT BOSS, DIRECT_REPORT, N

FROM FROM ORG_CHARTORG_CHARTGenerated

at Execution

SAM


MQT

MQT INDEX:BOSS

CREATE TABLE ORG_LIST AS (CREATE TABLE ORG_LIST AS (WITH ORG_CHART(BOSS, DIRECT_REPORT, N) ASWITH ORG_CHART(BOSS, DIRECT_REPORT, N) AS(SELECT BOSS, DIRECT_REPORT, 1(SELECT BOSS, DIRECT_REPORT, 1FROM FROM ORG_TABLEORG_TABLEWHEREWHERE BOSS= ‘SAM’BOSS= ‘SAM’UNION ALLUNION ALLSELECT SELECT T1.BOSS, T2.DIRECT_REPORT, N+1T1.BOSS, T2.DIRECT_REPORT, N+1FROM FROM ORG_CHART T2, ORG_TABLE T1ORG_CHART T2, ORG_TABLE T1WHEREWHERE T2.DIRECT_REPORT = T2.BOSS )T2.DIRECT_REPORT = T2.BOSS )SELECT BOSS, DIRECT_REPORT, NSELECT BOSS, DIRECT_REPORT, NFROM FROM ORG_CHART)ORG_CHART)

DATA INITIALLY DEFERREDDATA INITIALLY DEFERREDREFRESH DEFERREDREFRESH DEFERREDMAINTAINED BY SYSTEMMAINTAINED BY SYSTEMENABLE QUERY OPTMIZATION;ENABLE QUERY OPTMIZATION;


Query Speed Up

DIRECTBOSS _REPORT NJOHN ADAM 501BETH JIM 634BETH SUE 634BETH MARK 635BETH SAM 635

WITH ORG_CHART (BOSS, DIRECT_REPORT, N) AS(SELECT BOSS, DIRECT_REPORT, 1

FROM ORG_TABLEWHERE BOSS= ‘BETH’UNION ALLSELECT T1.BOSS, T2.DIRECT_REPORT, N+1FROM ORG_CHART T2, ORG_TABLE T1WHERE T2.DIRECT_REPORT = T2.BOSS AND N < 10 0,000)

SELECT BOSS, DIRECT_REPORT, NFROM ORG_CHART

Accessed at

Execution

SELECT * FROM ORG_LISTSELECT * FROM ORG_LISTWHERE WHERE BOSS = ‘BETH’ BOSS = ‘BETH’

ORG_LIST

Designing Indexes

For a WorkloadFor a Workload



Table/Column Usage TranTran c1c1 c2c2 c3c3 c4c4 c5c5 c6c6 c7c7 c8c8 c9c9 c10c10 c11c11tran1tran1

q1q1 OO11

GG11

=T4.C2=T4.C2

OO22

GG22

=:hv=:hv

OO33

GG33

>800>800

SS =T2.C1=T2.C1 SS SS SSBTW BTW 2and 2and 77

SUM SUM byby

C1.C2.C1.C2.C3C3

AVG byAVG by

C1.C2.C3C1.C2.C3

MAX MAX byby

C1.C2.C1.C2.C3C3

q2q2 ss ss ww =22=22 oo11=:hv=:hv oo22

q3q3 =T3.C1=T3.C1 ss ss ss ss ss oo11 ss =T3.C11=T3.C11 =T3.C1=T3.C1

tran2tran2

q1q1 OO11 OO22 OO33=:hv=:hv >:hv>:hv

q2q2 ss ss =:hv=:hv ss ss ss ss ss OO11 IN(list)IN(list) LIKE :LIKE :hvhv

O1 = Order By First PositionO1 = Order By First PositionG1 = Group By First PositionG1 = Group By First PositionS = SELECT listS = SELECT listW = Local or Join WHERE/ON or W = Local or Join WHERE/ON or Specify actual local/join filter Specify actual local/join filter Specify Group by Aggregates as: Specify Group by Aggregates as:

Aggr BY Column(s)Aggr BY Column(s)


Final Ranking

trantran11

100100

42004200 q1q1 OO1 1

GG11

WW

OO22

GG22

ww

OO33

GG33

ww

SS ww SS SS SS

ww

1010 q2q2 ss ss ww ww oo11 ww oo22

99009900 q3q3 ww oo11 ss ww ww

Weighting Based on Importance to the

Business

Frequency * Weight

420042004200

420042004200

10 10 10 10 10 10 10

9900

420042004200

420042004200

4200 4200 4200

9900 9900 9900 9900

WeightWeight

tran2tran2 300


Determining Weight Copy the frequency reportCopy the frequency report Sort by “business priorities”Sort by “business priorities”

» Use Work Load Manager Goals for guidanceUse Work Load Manager Goals for guidance» Use Business Unit Owner PrioritiesUse Business Unit Owner Priorities

Remove the top 30%Remove the top 30% Sort by FrequencySort by Frequency Pull of the top numberPull of the top number Multiply by 10Multiply by 10 Now you have the frequency number to beatNow you have the frequency number to beat

Let’s say 5,500 is top frequent

Then 55,000 is the goal

Pulls out most critical


Grand Total By Table, by OperationBy Table, by Operation

TotalTotal

ActivityActivity

c1c1 c2c2 c3c3 c4c4 c5c5 c6c6 c7c7 c8c8

OO1 1 5598055980

GG1 1 47904790

OO2 2 22202220

OO6 6 7070

W 3920W 3920

OO2 2 5598055980

GG2 2 47904790

W 400W 400

OO3 3 5598055980

GG3 3 47904790

W 3W 3

S S 660660

OO1 1

8740 8740

S 8740 S 8740

w w 2209022090

OO2 2

3940 3940

S 480S 480

S 4000S 4000

W 7640 W 7640

S 8390S 8390

W 2020W 2020

Use this information to justify making Use this information to justify making changes to current index designchanges to current index design


Summary Many Things Impact Query SpeedMany Things Impact Query Speed

• SQL exploitation and proper usSQL exploitation and proper us• Manual Query Rewrite when necessaryManual Query Rewrite when necessary• Optimal Index and MQT design Optimal Index and MQT design

It is Easy to Change the Optimizer’s MindIt is Easy to Change the Optimizer’s Mind The Trick … is to Know WHEN to use:The Trick … is to Know WHEN to use:

Extreme Tuning TechniquesExtreme Tuning TechniquesDISTINCT Table ExpressionsDISTINCT Table ExpressionsCorrelated Table ExpressionsCorrelated Table ExpressionsOR 0 = 1OR 0 = 1Anti-Join TechniqueAnti-Join TechniqueIndex and MQT DesignIndex and MQT DesignX2QBOptX2QBOpt

Traditional Tuning TechniquesTraditional Tuning TechniquesOPTIMIZE FOR n ROWSOPTIMIZE FOR n ROWSNo OpsNo OpsREOPT(VARS)REOPT(VARS)Fake FilteringFake FilteringIndex & MQT RedesignIndex & MQT Redesign


Recommended SQL Reading

DB2 Developer’s Guide, Sixth Edition SAMS Publishing, Craig S. MullinsCovers DB2 9 & 10 for z/OS!!!!!

DB2 for z/OS: Design Guidelines for High Performance and Availability http://www.redbooks.ibm.com/abstracts/sg247134.html

DB2 for z/OS Version 8 Performance Topics DB2 for z/OS Version 8 Performance Topics Explore the book online at Explore the book online at http://www.redbooks.ibm.com/abstracts/sg246465.html

DB2 9 for z/OS Performance Topics, SG24-7473-00, SG24-7473-00Explore the book online at Explore the book online at http://www.redbooks.ibm.com/abstracts/sg247473.html?Open

DB2 10 for z/OS Performance Topics DB2 10 for z/OS Performance Topics http://www.redbooks.ibm.com/abstracts/sg247942.html

Documents

IDUG 2013 Sheryl Larsen Tuning SQL