
- •Contents
- •Send Us Your Comments
- •Preface
- •Audience
- •Organization
- •Related Documentation
- •Conventions
- •Documentation Accessibility
- •1 Introduction to the Oracle Server
- •Database Structure and Space Management Overview
- •Logical Database Structures
- •Physical Database Structures
- •Data Dictionary Overview
- •Data Access Overview
- •SQL Overview
- •Objects Overview
- •PL/SQL Overview
- •Java Overview
- •XML Overview
- •Transactions Overview
- •Data Integrity Overview
- •SQL*Plus Overview
- •Memory Structure and Processes Overview
- •An Oracle Instance
- •Memory Structures
- •Process Architecture
- •The Program Interface Mechanism
- •An Example of How Oracle Works
- •Application Architecture Overview
- •Client/Server Architecture
- •Multitier Architecture: Application Servers
- •Distributed Databases Overview
- •Replication Overview
- •Streams Overview
- •Advanced Queuing Overview
- •Heterogeneous Services Overview
- •Data Concurrency and Consistency Overview
- •Concurrency
- •Read Consistency
- •Locking Mechanisms
- •Quiesce Database
- •Database Security Overview
- •Security Mechanisms
- •Database Administration Overview
- •Enterprise Manager Overview
- •Database Backup and Recovery Overview
- •Data Warehousing Overview
- •Differences Between Data Warehouse and OLTP Systems
- •Data Warehouse Architecture
- •Materialized Views
- •OLAP Overview
- •Change Data Capture Overview
- •High Availability Overview
- •Transparent Application Failover
- •Online Reorganization Architecture
- •Data Guard Overview
- •LogMiner Overview
- •Real Application Clusters
- •Real Application Clusters Guard
- •Content Management Overview
- •Oracle Internet File System Overview
- •2 Data Blocks, Extents, and Segments
- •Introduction to Data Blocks, Extents, and Segments
- •Data Blocks Overview
- •Data Block Format
- •Free Space Management
- •Extents Overview
- •When Extents Are Allocated
- •Determine the Number and Size of Extents
- •How Extents Are Allocated
- •When Extents Are Deallocated
- •Segments Overview
- •Introduction to Data Segments
- •Introduction to Index Segments
- •Introduction to Temporary Segments
- •Automatic Undo Management
- •3 Tablespaces, Datafiles, and Control Files
- •Introduction to Tablespaces, Datafiles, and Control Files
- •Oracle-Managed Files
- •Allocate More Space for a Database
- •Tablespaces Overview
- •The SYSTEM Tablespace
- •Undo Tablespaces
- •Default Temporary Tablespace
- •Using Multiple Tablespaces
- •Managing Space in Tablespaces
- •Multiple Block Sizes
- •Online and Offline Tablespaces
- •Read-Only Tablespaces
- •Temporary Tablespaces for Sort Operations
- •Transport of Tablespaces Between Databases
- •Datafiles Overview
- •Datafile Contents
- •Size of Datafiles
- •Offline Datafiles
- •Temporary Datafiles
- •Control Files Overview
- •Control File Contents
- •Multiplexed Control Files
- •4 The Data Dictionary
- •Introduction to the Data Dictionary
- •Structure of the Data Dictionary
- •SYS, Owner of the Data Dictionary
- •How the Data Dictionary Is Used
- •How Oracle Uses the Data Dictionary
- •How to Use the Data Dictionary
- •Dynamic Performance Tables
- •Database Object Metadata
- •Introduction to an Oracle Instance
- •The Instance and the Database
- •Connection with Administrator Privileges
- •Initialization Parameter Files
- •Instance and Database Startup
- •How an Instance Is Started
- •How a Database Is Mounted
- •What Happens When You Open a Database
- •Database and Instance Shutdown
- •Close a Database
- •Unmount a Database
- •Shut Down an Instance
- •6 Application Architecture
- •Client/Server Architecture
- •Multitier Architecture
- •Clients
- •Application Servers
- •Database Servers
- •Oracle Net Services
- •How Oracle Net Services Works
- •The Listener
- •7 Memory Architecture
- •Introduction to Oracle Memory Structures
- •System Global Area (SGA) Overview
- •Dynamic SGA
- •Database Buffer Cache
- •Redo Log Buffer
- •Shared Pool
- •Large Pool
- •Control of the SGA’s Use of Memory
- •Other SGA Initialization Parameters
- •Program Global Areas (PGA) Overview
- •Content of the PGA
- •SQL Work Areas
- •PGA Memory Management for Dedicated Mode
- •Dedicated and Shared Servers
- •Software Code Areas
- •8 Process Architecture
- •Introduction to Processes
- •Multiple-Process Oracle Systems
- •Types of Processes
- •User Processes Overview
- •Connections and Sessions
- •Oracle Processes Overview
- •Server Processes
- •Background Processes
- •Trace Files and the Alert Log
- •Shared Server Architecture
- •Scalability
- •Dispatcher Request and Response Queues
- •Shared Server Processes (Snnn)
- •Restricted Operations of the Shared Server
- •Dedicated Server Configuration
- •The Program Interface
- •Program Interface Structure
- •Program Interface Drivers
- •Communications Software for the Operating System
- •9 Database Resource Management
- •Introduction to the Database Resource Manager
- •Database Resource Manager Overview
- •Example of a Simple Resource Plan
- •How the Database Resource Manager Works
- •Resource Control
- •Database Integration
- •Performance Overhead
- •Resource Plans and Resource Consumer Groups
- •Activation of a Resource Plan
- •Groups of Resource Plans
- •Resource Allocation Methods and Resource Plan Directives
- •Resource Plan Directives
- •CPU Resource Allocation
- •Interaction with Operating-System Resource Control
- •Dynamic Reconfiguration
- •10 Schema Objects
- •Introduction to Schema Objects
- •Tables
- •How Table Data Is Stored
- •Nulls Indicate Absence of Value
- •Default Values for Columns
- •Partitioned Tables
- •Nested Tables
- •Temporary Tables
- •External Tables
- •Views
- •How Views are Stored
- •How Views Are Used
- •Mechanics of Views
- •Dependencies and Views
- •Updatable Join Views
- •Object Views
- •Inline Views
- •Materialized Views
- •Define Constraints on Views
- •Refresh Materialized Views
- •Materialized View Logs
- •Dimensions
- •The Sequence Generator
- •Synonyms
- •Indexes
- •Unique and Nonunique Indexes
- •Composite Indexes
- •Indexes and Keys
- •Indexes and Nulls
- •Function-Based Indexes
- •How Indexes Are Stored
- •How Indexes Are Searched
- •Key Compression
- •Reverse Key Indexes
- •Bitmap Indexes
- •Bitmap Join Indexes
- •Index-Organized Tables
- •Benefits of Index-Organized Tables
- •Index-Organized Tables with Row Overflow Area
- •Secondary Indexes on Index-Organized Tables
- •Bitmap Indexes on Index-Organized Tables
- •Partitioned Index-Organized Tables
- •Index-Organized Table Applications
- •Application Domain Indexes
- •Clusters
- •Hash Clusters
- •Introduction to Partitioning
- •Partition Key
- •Partitioned Tables
- •Partitioned Index-Organized Tables
- •Partitioning Methods
- •Range Partitioning
- •List Partitioning
- •Hash Partitioning
- •Composite Partitioning
- •When to Partition a Table
- •Partitioned Indexes
- •Local Partitioned Indexes
- •Global Partitioned Indexes
- •Global Nonpartitioned Indexes
- •Partitioned Index Examples
- •Miscellaneous Information about Creating Indexes on Partitioned Tables
- •Using Partitioned Indexes in OLTP Applications
- •Using Partitioned Indexes in Data Warehousing and DSS Applications
- •Partitioned Indexes on Composite Partitions
- •Partitioning to Improve Performance
- •Partition Pruning
- •Partition-wise Joins
- •Parallel DML
- •Introduction to Oracle Datatypes
- •Character Datatypes
- •CHAR Datatype
- •VARCHAR2 and VARCHAR Datatypes
- •Length Semantics for Character Datatypes
- •NCHAR and NVARCHAR2 Datatypes
- •Use of Unicode Data in an Oracle Database
- •LOB Character Datatypes
- •LONG Datatype
- •NUMBER Datatype
- •Internal Numeric Format
- •DATE Datatype
- •Use of Julian Dates
- •Date Arithmetic
- •Centuries and the Year 2000
- •Daylight Savings Support
- •Time Zones
- •LOB Datatypes
- •BLOB Datatype
- •CLOB and NCLOB Datatypes
- •BFILE Datatype
- •RAW and LONG RAW Datatypes
- •ROWID and UROWID Datatypes
- •The ROWID Pseudocolumn
- •Physical Rowids
- •Logical Rowids
- •Rowids in Non-Oracle Databases
- •ANSI, DB2, and SQL/DS Datatypes
- •XML Datatypes
- •XMLType Datatype
- •URI Datatypes
- •Data Conversion
- •13 Object Datatypes and Object Views
- •Introduction to Object Datatypes
- •Complex Data Models
- •Multimedia Datatypes
- •Object Datatype Categories
- •Object Types
- •Collection Types
- •Type Inheritance
- •FINAL and NOT FINAL Types
- •NOT INSTANTIABLE Types and Methods
- •User-Defined Aggregate Functions
- •Why Have User-Defined Aggregate Functions?
- •Creation and Use of UDAGs
- •How Do Aggregate Functions Work?
- •Application Interfaces
- •JPublisher
- •JDBC
- •SQLJ
- •Datatype Evolution
- •Introduction to Object Views
- •Advantages of Object Views
- •How Object Views Are Defined
- •Use of Object Views
- •Updates of Object Views
- •Updates of Nested Table Columns in Views
- •View Hierarchies
- •14 SQL, PL/SQL, and Java
- •SQL Overview
- •SQL Statements
- •Identification of Nonstandard SQL
- •Recursive SQL
- •Cursors
- •Shared SQL
- •Parsing
- •SQL Processing
- •The Optimizer Overview
- •PL/SQL Overview
- •How PL/SQL Runs
- •Language Constructs for PL/SQL
- •PL/SQL Program Units
- •PL/SQL Collections and Records
- •PL/SQL Server Pages
- •Java Overview
- •Java and Object-Oriented Programming Terminology
- •Class Hierarchy
- •Interfaces
- •Polymorphism
- •The Java Virtual Machine (JVM)
- •Why Use Java in Oracle?
- •Oracle’s Java Application Strategy
- •15 Dependencies Among Schema Objects
- •Introduction to Dependency Issues
- •Resolution of Schema Object Dependencies
- •Compilation of Views and PL/SQL Program Units
- •Function-Based Index Dependencies
- •Object Name Resolution
- •Shared SQL Dependency Management
- •Local and Remote Dependency Management
- •Management of Local Dependencies
- •Management of Remote Dependencies
- •16 Transaction Management
- •Introduction to Transactions
- •Statement Execution and Transaction Control
- •Statement-Level Rollback
- •Resumable Space Allocation
- •Transaction Management Overview
- •Commit Transactions
- •Rollback of Transactions
- •Savepoints In Transactions
- •Transaction Naming
- •The Two-Phase Commit Mechanism
- •Discrete Transaction Management
- •Autonomous Transactions
- •Autonomous PL/SQL Blocks
- •Transaction Control Statements in Autonomous Blocks
- •17 Triggers
- •Introduction to Triggers
- •How Triggers Are Used
- •Parts of a Trigger
- •The Triggering Event or Statement
- •Trigger Restriction
- •Trigger Action
- •Types of Triggers
- •Row Triggers and Statement Triggers
- •BEFORE and AFTER Triggers
- •INSTEAD OF Triggers
- •Triggers on System Events and User Events
- •Trigger Execution
- •The Execution Model for Triggers and Integrity Constraint Checking
- •Data Access for Triggers
- •Storage of PL/SQL Triggers
- •Execution of Triggers
- •Dependency Maintenance for Triggers
- •18 Parallel Execution of SQL Statements
- •Introduction to Parallel Execution
- •When to Implement Parallel Execution
- •When Not to Implement Parallel Execution
- •How Parallel Execution Works
- •Parallelized SQL Statements
- •Degree of Parallelism
- •SQL Operations That Can Be Parallelized
- •Parallel Query
- •Parallel DDL
- •Parallel DML
- •SQL*Loader
- •How to Make a Statement Run in Parallel
- •19 Direct-Path INSERT
- •Introduction to Direct-Path INSERT
- •Advantages of Direct-Path INSERT
- •Serial and Parallel Direct-Path INSERT
- •Direct-Path INSERT Into Partitioned and Nonpartitioned Tables
- •Serial Direct-Path INSERT into Partitioned and Nonpartitioned Tables
- •Parallel Direct-Path INSERT into Partitioned Tables
- •Parallel Direct-Path INSERT into Nonpartitioned Tables
- •Direct-Path INSERT and Logging Mode
- •Direct-Path INSERT with Logging
- •Direct-Path INSERT without Logging
- •Additional Considerations for Direct-Path INSERT
- •Index Maintenance with Direct-Path INSERT
- •Space Considerations with Direct-Path INSERT
- •Locking Considerations with Direct-Path INSERT
- •20 Data Concurrency and Consistency
- •Introduction to Data Concurrency and Consistency in a Multiuser Environment
- •Preventable Phenomena and Transaction Isolation Levels
- •Overview of Locking Mechanisms
- •How Oracle Manages Data Concurrency and Consistency
- •Multiversion Concurrency Control
- •Statement-Level Read Consistency
- •Transaction-Level Read Consistency
- •Read Consistency with Real Application Clusters
- •Oracle Isolation Levels
- •Comparison of Read Committed and Serializable Isolation
- •Choice of Isolation Level
- •How Oracle Locks Data
- •Transactions and Data Concurrency
- •Deadlocks
- •Types of Locks
- •DML Locks
- •DDL Locks
- •Latches and Internal Locks
- •Explicit (Manual) Data Locking
- •Oracle Lock Management Services
- •Flashback Query
- •Flashback Query Benefits
- •Some Uses of Flashback Query
- •21 Data Integrity
- •Introduction to Data Integrity
- •Types of Data Integrity
- •How Oracle Enforces Data Integrity
- •Introduction to Integrity Constraints
- •Advantages of Integrity Constraints
- •The Performance Cost of Integrity Constraints
- •Types of Integrity Constraints
- •NOT NULL Integrity Constraints
- •UNIQUE Key Integrity Constraints
- •PRIMARY KEY Integrity Constraints
- •Referential Integrity Constraints
- •CHECK Integrity Constraints
- •The Mechanisms of Constraint Checking
- •Default Column Values and Integrity Constraint Checking
- •Deferred Constraint Checking
- •Constraint Attributes
- •SET CONSTRAINTS Mode
- •Unique Constraints and Indexes
- •Constraint States
- •Constraint State Modification
- •22 Controlling Database Access
- •Introduction to Database Security
- •Schemas, Database Users, and Security Domains
- •User Authentication
- •Authentication by the Operating System
- •Authentication by the Network
- •Authentication by the Oracle Database
- •Multitier Authentication and Authorization
- •Authentication by the Secure Socket Layer Protocol
- •Authentication of Database Administrators
- •Oracle Internet Directory
- •User Tablespace Settings and Quotas
- •Default Tablespace Option
- •Temporary Tablespace Option
- •Tablespace Access and Quotas
- •The User Group PUBLIC
- •User Resource Limits and Profiles
- •Types of System Resources and Limits
- •Profiles
- •23 Privileges, Roles, and Security Policies
- •Introduction to Privileges
- •System Privileges
- •Schema Object Privileges
- •Table Security
- •View Security
- •Procedure Security
- •Type Security
- •Introduction to Roles
- •Common Uses for Roles
- •The Mechanisms of Roles
- •Grant and Revoke Roles
- •Who Can Grant or Revoke Roles?
- •Role Names
- •Security Domains of Roles and Users
- •PL/SQL Blocks and Roles
- •Data Definition Language Statements and Roles
- •Predefined Roles
- •The Operating System and Roles
- •Roles in a Distributed Environment
- •Fine-Grained Access Control
- •Dynamic Predicates
- •Application Context
- •Secure Application Roles
- •Creation of Secure Application Roles
- •24 Auditing
- •Introduction to Auditing
- •Features of Auditing
- •Mechanisms for Auditing
- •Statement Auditing
- •Privilege Auditing
- •Schema Object Auditing
- •Schema Object Audit Options for Views and Procedures
- •Fine-Grained Auditing
- •Focus Statement, Privilege, and Schema Object Auditing
- •Successful and Unsuccessful Statement Executions Auditing
- •BY SESSION and BY ACCESS Clauses of Audit Statement
- •Audit By User
- •Audit in a Multitier Environment
- •Allocating Extents in Dictionary Managed Tablespaces
- •Introduction to Rollback Segments
- •PCTFREE, PCTUSED, and Row Chaining
- •Glossary
- •Index

How Parallel Execution Works
Note: When a set of parallel execution servers completes its operation, it moves on to operations higher in the data flow. For example, in the previous diagram, if there was another ORDER BY operation after the ORDER BY, then the parallel execution servers performing the table scan perform the second ORDER BY operation after completing the table scan.
Degree of Parallelism
The parallel execution coordinator may enlist two or more of the instance’s parallel execution servers to process a SQL statement. The number of parallel execution servers associated with a single operation is known as the degree of parallelism.
Note that the degree of parallelism applies directly only to intra-operation parallelism. If inter-operation parallelism is possible, the total number of parallel execution servers for a statement can be twice the specified degree of parallelism. No more than two sets of parallel execution servers can run simultaneously. Each set of parallel execution servers may process multiple operations. Only two sets of parallel execution servers need to be active to guarantee optimal inter-operation parallelism.
Parallel execution is designed to effectively use multiple CPUs and disks to answer queries quickly. When multiple users use parallel execution at the same time, it is easy to quickly exhaust available CPU, memory, and disk resources.
Oracle provides several ways to manage resource utilization in conjunction with parallel execution environments, including:
The adaptive multiuser algorithm, which reduces the degree of parallelism as the load on the system increases. You can turn this option with the PARALLEL_ ADAPTIVE_MULTI_USER parameter of the ALTER SYSTEM statement or in your initialization parameter file.
User resource limits and profiles, which allow you to set limits on the amount of various system resources available to each user as part of a user’s security domain.
The Database Resource Manager, which lets you allocate resources to different groups of users.
18-8 Oracle9i Database Concepts

How Parallel Execution Works
See Also:
Oracle9i Database Reference for information about PARALLEL_ ADAPTIVE_MULTI_USER
Oracle9i SQL Reference for the syntax of the ALTER SYSTEM SQL statement
Oracle9i Data Warehousing Guide
Parallel Query Intraand Inter-Operation Example
As an example of parallel query with intraand inter-operation parallelism, consider a more complex query:
SELECT /*+ PARALLEL(employees 4) PARALLEL(departments 4) USE_HASH(employees) ORDERED */
MAX(salary), AVG(salary) FROM employees, departments
WHERE employees.department_id = departments.department_id GROUP BY employees.department_id;
Note that hints have been used in the query to force the join order and join method, and to specify the degree of parallelism (DOP) of the tables employees and departments. In general, you should let the optimizer determine the order and method.
The query plan or data flow graph corresponding to this query is illustrated in Figure 18–4.
Parallel Execution of SQL Statements 18-9

How Parallel Execution Works
Figure 18–4 Data Flow Diagram for Joining Tables
Parallel
Execution
Coordinator
GROUP
BY
SORT
HASH
JOIN
FULL SCAN employees
FULL SCAN departments
Given two sets of parallel execution servers SS1 and SS2, the execution of this plan will proceed as follows: each server set (SS1 and SS2) will have four execution processes because of the PARALLEL hint in the query that specifies the DOP. In other words, the DOP will be four because each set of parallel execution servers will have four processes.
18-10 Oracle9i Database Concepts

How Parallel Execution Works
Slave set SS1 first scans the table employees while SS2 will fetch rows from SS1 and build a hash table on the rows. In other words, the parent servers in SS2 and the child servers in SS2 work concurrently: one in scanning employees in parallel, the other in consuming rows sent to it from SS1 and building the hash table for the hash join in parallel. This is an example of inter-operation parallelism.
After SS1 has finished scanning the entire table employees (that is, all granules or task units for employees are exhausted), it scans the table departments in parallel. It sends its rows to servers in SS2, which then perform the probes to finish the hash-join in parallel. After SS1 is done scanning the table departments in parallel and sending the rows to SS2, it switches to performing the GROUP BY in parallel. This is how two server sets run concurrently to achieve inter-operation parallelism across various operators in the query tree while achieving intra-operation parallelism in executing each operation in parallel.
Another important aspect of parallel execution is the re-partitioning of rows while they are sent from servers in one server set to another. For the query plan in Figure 18–4, after a server process in SS1 scans a row of employees, which server process of SS2 should it send it to? The partitioning of rows flowing up the query
tree is decided by the operator into which the rows are flowing into. In this case, the partitioning of rows flowing up from SS1 performing the parallel scan of employees into SS2 performing the parallel hash-join is done by hash partitioning on the join column value. That is, a server process scanning employees computes a hash function of the value of the column employees.employee_id to decide the number of the server process in SS2 to send it to. The partitioning method used in parallel queries is explicitly shown in the EXPLAIN PLAN of the query. Note that the partitioning of rows being sent between sets of execution servers should not be confused with Oracle’s partitioning feature whereby tables can be partitioned using hash, range, and other methods.
See Also: Oracle9i Data Warehousing Guide for examples of using
EXPLAIN PLAN with parallel query
Parallel Execution of SQL Statements 18-11