SQL Tuning for Everyone JAI MAA DURGA

8/11/2019 SQL Tuning for Everyone JAI MAA DURGA

1/38

1


2/38

SQL Tuning for Smarties, Dummies and Everyone in Between

Novices

Jagan Athreya

Director, Database Manageability, Oracle

Arup Nanda

2

Senior Director, Database Architecture, Starwood Hotels and Resorts


3/38

The following is intended to outline our general

product direction. It is intended for information

purposes only, and may not be incorporated into any

material, code, or functionality, and should not be

relied upon in making purchasing decisions. , ,

features or functionality described for Oracles

products remains at the sole discretion of Oracle.

3


4/38

Outline

SQL Tuning Challenges

SQL Tuning Solutions New Feature Overview

Problem Root Causes and their Solutions

Preventing SQL Problems

Q & A

4


5/38


-

DBA team

Mostly average, some superstars

Superstars take most of the burden over-stretched

Development staff

Mostl non-Oracle skills Java, C++

Usually considers the DB as a black box

Writing efficient queries, troubleshooting performance issues

5


6/38


Situation:

Query from hell pops up

Brings the database to its knees

Response

DBA: Developer should be taking care of this.

Developer: Why is the DBA not aware of this problem?

Manager: DBA will review all queries and approve them.

What is the most efficient way to manage this process?

6


7/38


Situation

New SQL statements added as part of application patch

deployment

Database u rades

Database patching

Response Users: How will the application perform after the changes?

DBA: How do I ensure that our SLA remains intact after the

changes are rolled out? Challenge

How to reduce business risk while absorbing new

7


8/38


Situation

Data in Production has evolved over time. Have the optimizer

statistics stayed current?

DBA:

Will statistics refresh break something?

What will happen if we dont collect?

How often should I collect the statistics ?

What happens when you collect a new set? Challenge

What is the recommended strategy for managing optimizer

8


9/38


No time to find the root cause. How to prevent this

from recurring?

Bind variables: How do you prevent bad plans based

How to diagnose a bad plan

10053 trace, endless ourin over traces

Wrongly constructed predicates

How to fix a bad plan

n s c ange o co e

Baselines vs. SQL Profiles

Pick out a single SQL or a bunch from the shared pool

9


10/38

Outline





Q & A

10


11/38

Real-Time SQL Monitoring

Lookin Inside SQL Execution

Automatically monitors longrunning SQL

Enabled out-of-the-box with noperformance overhead

Monitors each SQL execution

Exposes monitoring statistics

Global execution level

Plan operation level

Parallel Execution level

Guides tuning efforts

11


12/38

New capabilities in SQL Monitoring

New in Oracle Database 11 Release 2

PL/SQL monitoring including associated high load SQL monitored recursively

Exadata aware I/O erformance monitorin and associated metric data

Capture rich metadata such as bind values, session details e.g. user,program, client_id and error codes and error messages

Save as Active Report for rich interactive offline analysis

12


13/38

DEMO

13


14/38

Application Tuning

High-Load

Packaged Apps +SQL Profi le

Well-Tuned SQL

us om za e pps +SQL Advice

Customizable Apps +Indexes & MVs +Partitions

Appl ications

Automatic Tuning Opt imizer

Automatic SQL Tuning

Identifies high-load SQL from AWR

Tunes SQL usin SQL Profiles Implements greatly improved SQL plans (optional)

Performance benefit of advice provided

SQL Profiling tunes execution plan without changing SQL text

14

Enables transparent tuning for packaged applications


15/38

Automatic SQL Tuning

Gather Missing or Stale

Statistics

SQL Profiling

Statistics Analysis

Access Path Analysis

SQL Restructure Analysis

rea e a ro e

Add Missing Access

Structures

Modify SQL Constructs

Al ternative Plan Analysis

Parallel Query AnalysisAdministrator

Comprehensive

op erna ve

Execution Plan

Create Parallel SQL

Profile

SQL TuningAdvisor

SQL Tuning Advisor

-

OptimizerSQL Tuning

Recommendations

performance data

NEW: Recommends parallel profile if it will improve SQL performance

significantly (2x or more)

15


16/38

SQL Tuning for DevelopersIntegration with Visual Studio

Introduced in Oracle Developer Tools for Visual Studio Release 11.1.0.7.20 Oracle Performance Analyzer

Tune running applications with the help of ADDM

Query Window

Tune individual SQL statements with STA Server Explorer

16


17/38

Agenda





Q & A

17

What makes SQL go bad?


18/38

What makes SQL go bad?Root Causes of Poor SQL Performance

1. Optimizer statistics issues

a. Stale/Missing statistics

b. Incomplete statistics

c. Improper optimizer configuration

d. U raded database: new o timizer

e. Changing statistics

f. Rapidly changing data2. Application Issues

a. ss ng access s ruc ures

b. Poorly written SQL statements

3. Cursor sharing issuesa. Bind-sensitive SQL with bind eekinb. Literal usage

4. Resource and contention issues

a. Hardware resource crunch

. on en on row oc con en on, oc up a e con en on

c. Data fragmentation

5. Parallelism issues

a. Not parallelized (no scaling to large data)

18

b. Improperly parallelized (partially parallelized, skews)

Wh t k SQL b d?


19/38


1. Optimizer statistics issuesa. Stale/Missing statistics

b. Incomplete statistics

c. Improper optimizer configuration

d. Upgraded database: new optimizer

e. Changing statistics

f. Rapidly changing data

2. Application Issues

3. Cursor sharing issues



19


20/38

Oracle Optimizer Statistics

Inaccurate statistics Suboptimal Plans

Table Statistics

Column Statistics

Index Statistics

20


21/38

Oracle Optimizer StatisticsPreventing SQL Regressions

NoviceMode

Automatic Statistics Collection Job (stale ormissin

Out-of-the box, runs in maintenance window

Configuration can be changed (at tablelevel)

a ers s a s cs on user an c onaryobjects

Uses new collection algorithm with Nightly

sampling of 10%

Incrementally maintains statistics for

Set DBMS_STATS.SET_GLOBAL_PREFS

21


22/38

Oracle Optimizer StatisticsPreventing SQL Regressions

ExpertMode

Extended Statistics Extended Optimizer Statistics provides a mechanism

to collect statistics on a group of related columns:

Function-Based Statistics Multi-Column Statistics

Automatically maintained with column statisticsDBMS_STATS. CREATE_EXTENDED_STATS

Allows validation of statistics before publishing

Disabled by default

To enable set table/schema PUBLISH settin to FALSEDBMS_STATS. SET_TABLE_PREFS( ' SH' , ' CUSTOMERS' , ' PUBLI SH' , ' f al se' )

To use for validation

ALTER SESSI ON SET opt i mi zer _pendi ng_st at i st i cs = TRUE;

22

Publish after successful verification



23/38


1. Optimizer statistics issues


.

b. Poorly written SQL statements


.


23


24/38

Identify performance problems using ADDMAutomatic Database Diagnostic Monitor

Novice

Provides database and

o e

diagnostic

Throughput centric - Focus

on reducing time DB time

Identifies top SQL:

Shows SQL impact

Frequency of

occurrence

Pinpoints root cause:

SQL stmts waiting for

SQL stmts not shared

24


25/38

Identify High Load SQL Using Top ActivityNovice

Performance Page

Identify Top SQL by DB Time:

CPU

o e

I/O

Non-idle waits

Different Levels of Analysis

Historical analysis

AWR data

Performance Page

Top Activity

Real-time analysis

ASH data

More ranular anal sis Enables identification of

transient problem SQL

Top Activity Page

25

Tune using SQL Tuning Advisor


26/38

Advanced SQL TuningUniverse of Access Structures

Novice+Mode

Indexes: B-tree indexes, B-tree cluster indexes, Hash cluster indexes, Globaland local indexes, Reverse key indexes, Bitmap indexes, Function-basedindexes, Domain indexes

Materialized Views:Primary Key materialized views,

ROWID materialized views

Complex materialized views

Partitioned Tables:

Range partitioning, Hash partitioning,

List partitioning, Composite partitioning,

n erva ar on ng, par on ng,

Virtual Column Based partitioning

B-tree index

26


27/38

SQL Access Advisor: Partition AdvisorNovice+

Mode

Indexes

Materialized

viewsSQL AccessAdvisor

RepresentativeWorkload

Materialized

views logs

Automatic Tuning

Optimizer

Partitioned

ob ects

Analysis

27


28/38

SQL Access AdvisorAdvanced Options

ExpertMode

Workload filtering Limited vs. advanced mode

Hypothetical workload tuning

Factoring in the cost of creation

28



29/38


. p m zer s a s cs ssues



.

b. Bind-sensitive SQL with bind peeking


29


30/38

What makes SQL go bad?a. Literal Usage Issue

ExpertMode

SELECT * FROM jobs WHERE min_salary > 12000;



SELECT * FROM

SELECT * FROM *

cursor_sharing ={exact, force, similar}

ar ngCursorsis good!

30


31/38

What makes SQL go bad?b. Bind Peeking Issue

NMode

Processed_FlagY

Full Table Scan

Y

Y CBO10gFTS

99

Two different optimal plansfor different bind values

Index Range Scan

N

IRS

1

31

Problem: Binds will affect optimality in any subsequent uses of the stored plan

Fi i bl ith Ad ti C Sh i


32/38

Fixing problems with Adaptive Cursor SharingAdaptive Cursor Sharing Expert

SELECT * FROM emp WHERE wage := wage_value

1

2

20 25

SamePlan

3

22 24

DifferentPlan

4

30 35

SamePlan,

ExpandInterval

32

34 43


33/38

Agenda





Q & A

33

P ti bl ith SQL Pl M t


34/38

Preventing problems with SQL Plan Management

Problem: changes in the

environment cause plans to changeNL

GB

Parse

Plan baseline is establishedNL

Statement log

Plan history

SQL statement is parsed again and

a different plan is generated

HJ

GB

Plan baselineGB

NL

NL

New plan is not executed butHJmar e or ver ca on

34

SQL Plan Management


35/38

SQL Plan ManagementMigration of Stored Outlines to Plan Baselines

Oracle Database 11g

GB 5. Migrate Stored OutlinesPlan History

GB

OH Schema

No plan

regressions

HJ

HJ

HJ

HJHJ

HJ

4. Upgrade

to 11g

Oracle Database 11g

OH Schema

HJ

GB

CREATE_STORED_OUTLINES=true. eg n

with

2. Run all SQL in the

Application and auto

Oracle Database 9 or 10g

HJ

CREATE_STORED_OUTLINES=false

crea e a ore u ne

for each one

3. After Store

Outlines are

ca tured

35

SQL Performance Analyzer (SPA)


36/38

SQL Performance Analyzer (SPA)Validate statistics refresh with SPA

Steps:

1. Capture SQL workload in STS

Validating upgrade with SPASQL Workload

capture capability

2. Execute SPA pre-change trial3. Refresh statistics using

SQL plans + stats SQL plans + stats

opt on

4. Execute SPA post-change trial

5. Run SPA report comparing SQL Compare

- -

Before PUBLISHing stats:

Remediate individual few SQLAnalysis Report

SQL Performance

for plan regressions: SPM, STA

Revert to old statistics if toomany regressions observed

36

C l i


37/38

ConclusionIdentify, Resolve, Prevent

PreventSPASPM

1. Production Performance

2. Change Causing Problems

3. Optimizer Statistics Management

. a p ans agnos s an eso u on

ADDM, Top Activity, SQL Monitoring

Identify

,

Access Advisor,

Auto Stat Collect ion

Top Activity,

ADDM,

SQL Monitoring

37


38/38

Documents

SQL Tuning for Everyone JAI MAA DURGA