Oracle Performance Survival Guide

Guy Harrisonhas worked with Oracle databases as a developer, administrator, and performance expert for more than two decades. He is the author of many articles and several books on database technology, including Oracle SQL High Performance Tuning(Prentice Hall) and MySql Stored Procedure Programming(with Steven Feuerstein, O’Reilly). Guy is currently a director of development at Quest Software and is the chief architect of Quest’s popular Spotlight product family. He lives in Melbourne, Australia, with his wife Jenni and children Chris, Kate, Mike, and William. You can find Guy on the Web at http://www.guyharrison.net

Preface

Part I: Methods, Concepts, and Tools

Chapter 1. Oracle Performance Tuning: A Methodical Approach

A Brief History of Oracle Performance Tuning

Moving Beyond a Symptomatic Approach

Stage 1: Minimizing the Application Workload

Stage 2: Reducing Contention and Bottlenecks

Stage 3: Reducing Physical IO

Stage 4: Optimizing Disk IO

Summary

Chapter 2. Oracle Architecture and Concepts

The Oracle APIs

    Creating the Cursor

    Checking for Cached SQL Statements 

    Parsing the SQL

    Associating Bind Variables

    Executing the SQL

    Fetching Rows

    Using Array Fetch

    Processing Result Sets

    Closing the Cursor

    Optimizing Oracle API Calls

The Oracle Query Optimizer

    Cost Based Optimization

    Optimizer Goal

    Optimizer Statistics

    Bind Variable Peeking and Adaptive Cursor Sharing

    Hints

    Outlines, Profiles, and Baselines

Transactions and Locking

Oracle Server Architecture

    Instances and Databases

    The System Global Area

    Data Caching

    The Program Global Area

    Memory Management

Segments and Files

    Tables

    Indexes

    Blocks, Extents, Segments, and Partitions

    Tablespaces and Data Files

    Undo Segments

    Redo Logs and Archive Logs

    Flashback Logs

Server Processes

Background Processes

Real Application Clusters

Summary

Chapter 3. Tools of the Trade

Explaining SQL Statements

    The Plan Table

    Exploiting Cached SQL

    DBMS_XPLAN

    Interpreting the Execution Plan

    Virtual Indexing

Tracing Oracle Execution

    Tracing from Within Your Session

    Identifying Your Trace File

    Getting Tracing Status

    Invoking Trace in Another Session

    Tracing by MODULE, ACTION, or SERVICE

    Starting a Trace Using a Login Trigger

    Finding the Trace File

    Other Specialized Traces

Formatting Traces with tkprof

    The tkprof Sort Options

    Other tkprof Options

    Merging Multiple SQL Trace Files

    Interpreting Tkprof Output

    Execution Plans in tkprof

    Wait Statistics and tkprof

    Alternatives to tkprof

Using AUTOTRACE in SQL*PLUS

Monitoring the Oracle Server

    The V$ table interface

    Wait Interface

    The Time Model

    Integrating the Time Model and Wait Interface

    Oracle Enterprise Manager

    Spotlight on Oracle

Summary

Part II: Application and Database Design

Chapter 4. Logical and Physical Database Design

Logical Data Modeling

    Normalization and Third Normal Form

    Data Type Choices

    Artificial Keys

    Data Warehouse Design

Logical to Physical

    Mapping Entities or Classes to Tables

    Choosing a Table Type

    Data Types and Precisions

    Optional Attributes and NULL Values

    Column Order

    Exploiting Oracle Object Types

Denormalization

    Replicating Column Values to Avoid Joins

    Summary Tables

    Vertical Partitioning

    Implementing Denormalization

Star Schema Design

    Star Schema Basics

    Snowflakes Schemas

    Dimension Hierarchies

    Aggregations and Materialized Views

    Materialized View Best Practices

Physical Storage Options

    Manual and Automatic Segment Storage Management

    Concurrent Inserts and Freelists

    PCTFREE and PCTUSED

    Compression

    LOB Storage

Oracle Partitioning

    Types of Partitions

    Composite Partitions

    Choosing a Partitioning Strategy

    Enterprise Manager Partitioning Advisor

Summary

Chapter 5. Indexing and Clustering

Overview of Oracle Indexing and Clustering

B*-Tree Indexes

    Index Selectivity

    Unique Indexes

    Implicit Indexes

    Concatenated Indexes

    Index Skip Scans

    Guidelines for Concatenated Indexes

    Index Merges

    Null Values in Indexes

    Reverse Key Indexes

    Index Compression

    Functional Indexes

    Foreign Keys and Locking

    Indexes and Partitioning

Bitmap Indexes

    Features of Bitmap Indexes

    Drawbacks of Bitmap Indexes

    Bitmap Indexes and Cardinality

    Bitmap Index Merge

    Bitmap Join Indexes

Index Overhead

    Index Organized Tables

    Configuring the Overflow Segment

    Periodic Rebuild of Index Only Tables

Clustering

    Index Clusters

    Hash Clusters

Nested Tables

Choosing the Best Indexing Strategy

Summary

Chapter 6. Application Design and Implementation

SQL Statement Management

    Optimizing Parsing

    Avoiding Unnecessary SQL Executions

The Array Interface

    Implementing Array Fetch

    Array Insert

Transaction Design

    Isolation Levels

    Transactions and Locks

    Row Level Locking in Oracle

    Application Locking Strategies

Using Stored Procedures to Reduce Network Traffic

Summary

Part III: SQL and PL/SQL Tuning

Chapter 7. Optimizing the Optimizer

The Oracle Optimizer

    What Is Cost?

    Optimizer Goal

    Selectivity and Cardinality

    Query Transformation

    Cost Calculations

    Object Statistics

    Histograms

    Bind Variable Peeking

    Adaptive Cursor Sharing

    Database Parameters

    System Statistics

    Collecting Statistics

Using DBMS_STAT

    DBMS_STATS Procedures and Parameters

    Setting DBMS_STATS Defaults

    Creating Histograms with METHOD_OPT

    Sampling

    Partition Statistics

    Extended Statistics

    Locking Statistics

    System Statistics

    Exporting and Importing Statistics

    Manipulating Statistics

Summary

Chapter 8. Execution Plan Management

Hints

    Using Hints to Change the Access Path

    Using Hints to Change the Join Order

    Errors in Hint Specifications

Stored Outlines

    Creating an Outline to Stabilize a Plan

    Hacking an Outline

    SQL Tuning Sets

    Manually Creating a Tuning Set

    Creating Tuning Sets in

    Enterprise Manager

SQL Profiles and the SQL Tuning Advisor

    Using DBMS_SQLTUNE

    Indexing Advice

    SQL Tuning in Enterprise Manager

    Cross-SQL Tuning with the SQL Access Advisor

SQL Baselines

    Creating the Baseline

    Evolving the Baseline

    Automating and Configuring Baselines

    Fixed Baselines

    Baseline Management in Oracle

    Enterprise Manager

Summary

Chapter 9. Tuning Table Access

Single Value Lookups

    Choosing Between Table and Index Scan

    Bitmap Indexes and Single Value Lookups

    Hash Clusters and Single Value Lookups

Avoiding “Accidental” Table Scans

    NOT EQUALS Conditions

    Searching for Nulls

    Searching for Values That Are NOT NULL

    Creating Indexes on NULLable Columns

    Unintentionally Disabling an Index with a Function

    Functional Indexes

    Functional Indexes and Statistics

    Virtual Columns

Multicolumn Lookups

    Using Concatenated Indexes

    Index Merges

    Uniqueness and Over-Indexing

Searching for Ranges

    Unbounded Range Scan

    Bounded Range Scans

    Range Lookups

Using the LIKE Operator

Multvalue Single-Column Lookups

Optimizing Necessary Full Table Scans

    Lowering the High Water Mark

    Optimizing PCTFREE and PCTUSED

    Reducing the Row Length

    Compressing the Table

    Making Database IO More Efficient

    Using the SAMPLE Option

    Parallel Query

    The Fast Full Index Scan

    Partitioning

Summary

Chapter 10. Joins and Subqueries

Types of Joins

Join Methods

    Nested Loops Join

    Sort-Merge Join

    Hash Join

Choosing the Right Join Method

    Sort-Merge/Hash Versus Nested Loops

    Sort-Merge Versus Hash Joins

Optimizing Joins

    Optimizing Nested Loops Join

    Optimizing Sort-Merge and Hash Joins

Avoiding Joins

    Denormalization

    Index Clusters

    Materialized Views

    Bitmap Join Index

Join Order

    Special Joins

    Outer Joins

    Star Joins

    Hierarchical Joins

Subqueries

    Simple Subqueries

    Correlated Subqueries

    Anti-Join Subqueries

    Semi-Join Subqueries

Summary

Chapter 11. Sorting, Grouping, and Set Operations

Sort Operations

    Optimal, One-Pass and Multi-Pass Sorts

    Measuring Sort Activity

    Tracing Sort Activity

    Using an Index to Avoid a Sort

Grouping and Aggregates

    Aggregate Operations

    Maximums and Minimums

    The “Top N” Query

    Counting the Rows in a Table

    GROUP BY Operations

    HAVING Versus WHERE

SET Operations

    UNION Versus UNION ALL

    INTERSECT

    MINUS

    SET Operations and Their Alternatives

Summary

Chapter 12. Using and Tuning PL/SQL

Performance Advantages of PL/SQL

    A Procedural Approach

    Reduction in Network Overhead

    Divide and Conquer Massive SQLs

Measuring PL/SQL Performance

    Measuring PL/SQL Overhead

    Using DBMS_PROFILER

    The 11g Hierarchical Profiler

    Data Access Optimization

    Array Processing and BULK COLLECT

    Array Processing for INSERT Statements

    Bind Variables and Dynamic SQL

PL/SQL Code Optimization

    Tune the SQL First

    PLSQL_OPTIMIZE_LEVEL

    LOOP Optimization

    “Short Circuiting” Expressions

    Order of Expressions in IF and CASE Statements

    Recursion

    The NOCOPY Clause

    Associative Arrays

Other Optimizations

    Native Compilation

    PL/SQL In-Lining

    Data Types

    Using Java for Computation

    Function Caching

DML Trigger Performance

    UPDATE OF and WHEN Clauses

    Before and After Row Triggers

Summary

Chapter 13. Parallel SQL

Understanding Parallel SQL

    Parallel Processes and the Degree of Parallelism

    Parallel Slave Pool

    Parallel Query IO

    Parallel Performance Gains

Deciding When to Use Parallel Processing

    Your Server Computer Has Multiple CPUs

    The Data to Be Accessed Is on Multiple Disk Drives

    The SQL to Be Parallelized is Long Running or Resource-Intensive

    The SQL Performs at Least One Full Table, Index, or Partition Scan

    There Is Spare Capacity on Your Host

    The SQL is Well Tuned

Configuring Parallel Processing

    Determining the Degree of Parallelism

    Parallel Hints

    Parallel Configuration Parameters

Monitoring Parallel SQL

    Parallel Explain Plans

    Tracing Parallel Execution

    The V$PQ_TQSTAT View

    Other Statistics

Optimizing Parallel Performance

    Start with a SQL That Is Optimized for Serial Execution

    Ensure That the SQL Is a Suitable SQL for Parallel Execution

    Ensure That the System Is Suitably Configured for Parallel Execution

    Make Sure that All Parts of the Execution Plan Are Parallelized

    Ensure That the Requested DOP Is Realistic

    Monitor the Actual DOP

    Check for Skew in Data and Skew in Workload Between Processes

Other Parallel Topics

    Parallel Execution in RAC

    Parallel Index Lookups

    Parallel DML

    Parallel DDL

Summary

Chapter 14. DML Tuning

DML Performance Fundamentals

    WHERE Clause Optimization

    Index Overhead

    Trigger Overhead

    Referential Integrity

INSERT Specific Optimizations

    Array Processing

    Direct Path Inserts

    Multi-Table Insert

    Manual Segment Storage Management (MSSM) and Freelists

    Parallel DML

DELETE Operations

    TRUNCATE

    Partitions

    Create Table as Select

UPDATE and MERGE Operations

    Correlated UPDATEs

    Optimizing MERGE

COMMIT Optimization

    COMMIT Frequency

    Batch and NOWAIT Commit

    NOLOGGING

Summary

Part IV: Minimizing Contention

Chapter 15. Lock Contention

Lock Types and Modes

Waiting for Locks

Monitoring and Analyzing Locks

    Lock Wait Statistics

    Finding the Responsible SQL

    Measuring Lock Contention for Specific Transactions

    Tracing Lock Activity

    Blockers and Waiters

Application Locking Strategies

When Row Level Locking Fails

    Unindexed Foreign Keys

    ITL Waits

    Bitmap Indexes

    Direct Path Inserts

System Locks

    The High Water Mark (HW) Enqueue

    The Space Transaction (ST) Enqueue

    The Sequence Cache (SQ) Enqueue

    The User Lock (UL) Enqueue

    Other System Locks

Summary

Chapter 16. Latch and Mutex Contention

Overview of Latch and Mutex Architecture

    Gets, Spins, and Sleeps

    Mutexes

Measuring and Diagnosing Latch/Mutex Contention

    Identifying Individual Latches

    Finding SQLs and Segments Associated with Latch Waits

Specific Latch/Mutex Scenarios

    Library Cache Mutex Waits

    Library Cache Pin

    Shared Pool Latch

    Cache Buffers Chains Latch

    Row Cache Objects Latch

    Other Latch Scenarios

Is Latch Contention Inevitable?

    What About Changing _SPIN_COUNT?

    Spin Count, Latch Contention, and Throughput

    Setting Spin Count for Individual Latches

Summary

Chapter 17. Shared Memory Contention

Buffer Cache Architecture

Free Buffer Waits

    DBWR Direct and Asynchronous IO

    Other Remedies for Free Buffer Waits

Recovery Writer (RVWR) Waits

    Improving Flashback Log IO

    Increasing the Size of the

    Flashback Log Buffer

Buffer Busy Waits

    Measuring Buffer Busy

    Traditional Causes of Buffer Busy Waits

    Buffer Busy and Hot Blocks

Redo Log Buffer Waits

Summary

Part V: Optimizing Memory

Chapter 18. Buffer Cache Tuning

Buffer Cache Principles

    The LRU List

    Table Scan Handling

    The CACHE Property

    Direct Path IO

Buffer Cache Configuration and Tuning

    Monitoring the Buffer Cache

    The Buffer Cache Hit Rate

    Multiple Buffer Caches

    Sizing the Buffer Cache

Automatic Shared Memory Management (ASMM)

    Implementing ASMM

    Monitoring Resize Operations

    Tuning ASMM

    Nondefault Pools

    Memory Thrashing

Summary

Chapter 19. Optimizing PGA Memory

IO and PGA Memory

PGA Memory Management

    PGA_AGGREGATE_TARGET

    Session PGA Limits

Measuring PGA Usage and Efficiency

    Session PGA Utilization

    Measuring Temporary IO Wait Time

    Measuring Work Area Activity

Sizing the PGA with V$PGA_TARGET_ADVICE

Over-Riding PGA Aggregate Target

Summary

Chapter 20. Other Memory Management Topics

Optimizing Overall Oracle Memory

    IO Wait Times and Memory Optimization

    Using Advisories to Distribute PGA/Buffer Cache Memory

    Oracle 11G Automatic Memory Management (AMM)

Result Set Cache

    Enabling and Configuring the Result Set Cache

    Result Cache Statistics

    Result Cache Dependencies

    Result Cache Latches

    PL/SQL Function Cache

Other Memory Optimizations

    Sizing the Shared Pool

    Large Pool Sizing

    Redo Log Buffer

    Locking the SGA

Summary

Part VI: IO Tuning and Clustering

Chapter 21. Disk IO Tuning Fundamentals

Disk IO Concepts

    Service Time and Throughput

    Queuing

    Disk Drives: Slow and Getting Slower

    Disk Capacity and Data Placement

Oracle IO Architecture

    Datafile Single Block Read

    Multi Block Read

    Direct Path Reads

    Temporary Direct Path IO

    Data File Write IO

    Direct Path Writes

    Redo Log IO

    Archive Log IO

    Flashback IO

    Control File IO

Measuring and Monitoring Oracle IO

    IO Wait Times

    Monitoring Datafile IO

    Calibrating IO

Optimizing Datafile IO

    Minimizing IO Latency

    Maximizing IO Throughput

    Striping Strategies

    RAID Arrays

    Isolating Datafile IO

Redo and Archive Optimization

    Alternating and Distributing Logs

    Redo and Archive Fine-Grained Striping

    Just Say NO to RAID5 for Redo!

    Redo Log Sizing

Flashback Logs

Summary

Chapter 22. Advanced IO Techniques

Automatic Storage Management (ASM)

    ASM Architecture

    ASM Monitoring

    ASM Tuning

Solid State Disk (SSD)

    Flash-Based SSD

    DDR RAM-Based SSD

    Hybrid SSD

    Using SSD for Oracle Databases

The Exadata Storage Server

Database Block Size

Summary

Chapter 23. Optimizing RAC

RAC Overview

    Global Cache Requests

    RAC Tuning Principles

    Single Instance Tuning and RAC  

Measuring Cluster Overhead

Reducing Global Cache Latency

    Measuring Global Cache Latency

    Examining the Interconnect

    Signs of Interconnect Problems

Optimizing the Interconnect

    Network Hardware and Protocols

    Ethernet Jumbo Frames

    UDP Buffer Size

    LMS Waits

Cluster Balance

    Assessing Cluster Balance

    Cluster Balance and Services

    RAC Load Balancing Facilities

Minimizing Global Cache Requests

    Causes of High Global Cache Request Rates

    Measuring Global Cache Request Rates

    Techniques for Reducing Global Cache Requests

Summary

Bibliography

 

 

9780137011957   TOC   9/21/2009