Sei sulla pagina 1di 68

This presentation is for informational purposes only and may not be incorporated into a contract or agreement.

<Insert Picture Here>

Partitioning and Purging Best Practices for the E-Business Suite


Ahmed Alomari & Mohsin Sameen Applications Performance Group

The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decision. The development, release, and timing of any features or functionality described for Oracles products remains at the sole discretion of Oracle.

Agenda
Partitioning Best Practices
Partitioning Concepts Supportability Partitioning Methods Customer Case Studies

Purging
Business Processes Purging Strategies New Purge Portal

Q&A

<Insert Picture Here>

Partitioning in the E-Business Suite

Partitioning Continues to Evolve

Database Release
Oracle 8

Partitioning Functionality
Range partitioned tables Local partitioned indexes Global range partitioned indexes Hash and Composite; Range-hash partitioned tables, Range partitioned index-organized tables List partitioned tables, Hash partitioned indexorganized tables Composite range-list partitioned tables List partitioned index-organized tables Global hash partitioned indexes

Oracle 8i Oracle 9i Oracle 9iR2 Oracle 10g

Partitioning Concepts
Data Partitioning:
Process of logically and/or physically segmenting data (and its associated storage, tables and indexes) into more manageable pieces, in order to improve manageability and accessibility. Example Divide large Applications tables and their indexes such as GL_BALANCES into smaller segments.

Partitioning in the eBusiness Suite


Oracle Applications utilizes partitioning in the standard product in many modules:
Advanced Planning and Scheduling Payables (Trial Balances) Projects Resources Workflow

Directory Services Runtime tables


Daily Business Intelligence HR (Employee Directory) Engineering

Support Policy: Custom Partitioning


Oracle Applications fully supports the use of custom partitioning of either Applications standard or custom tables. Definition of custom partitioning: Changing the partitioning definition of an existing Applications table as delivered (out-of-the-box). If custom partitioning is used, you must license the partitioning option. Refer to the write-up on the Technology Stack blog: http://blogs.oracle.com/schan/2006/09/06#a663

<Insert Picture Here>

Why Use Partitioning?

Partitioning Advantages
Table availability
Significantly reduce recovery times of key transaction tables by recovering the current partitions first.

Table manageability
Backup, restore, and rebuild at the partition level. Index rebuilds can be performed at the partition level. Partition aware operations such as MOVE, EXCHANGE, REBUILD can be used without affecting active partitions.

Partitioning Advantages
Performance
Improves access path of most queries since the majority of the access involves current data as opposed to historical data. Optimizer automatically prunes unnecessary partitions. Analytical reports or period close jobs/reports improve by scanning the current partition as opposed to all the partitions. Improves purge performance. Significantly improves upgrade performance Minimizes upgrade downtime.

Partitioning Advantages
CBO is partition aware and employs partition pruning at runtime.
Eliminates partitions which are not needed by the SQL statement. Joins are also partition aware and benefit from pruning.

Parallel operations can occur at the partition level. Utilizes the Information Lifecycle Management (ILM) infrastructure to allow partitions containing historical Applications data to be placed on lower cost storage devices.

<Insert Picture Here>

Partitioning Methods

Partitioning Methods The following partitioning methods are available:

Range partitioning

Hash partitioning

Composite partitioning

List partitioning

Partitioning Methods - Range


Range partitioning requires that rows identified by a "partition key" fall into a predefined range of values. Value of the partition key determines the partition to which a row belongs. Range partitioning was introduced in Oracle8. Range partitioning is useful for Applications tables for which the partition key matches the access path key.

Partitioning Methods - Range


 Partition the GL runtime tables by period_name
GL_BALANCES
476,720,210 rows

JUN-04 4,886,548 rows

JUL-04 4,949,399 rows

AUG-04 3,833,931 rows

SEP-04 464,417 rows

OCT-04 34,212 rows

Partitioning Methods
Range Partitioning Example:
CREATE TABLE GL_BALANCES (SET_OF_BOOKS_ID NUMBER(15) NOT NULL, CODE_COMBINATION_ID NUMBER(15) NOT NULL, CURRENCY_CODE VARCHAR2(15) NOT NULL, PERIOD_NAME VARCHAR2(15) NOT NULL, ACTUAL_FLAG VARCHAR2(1) NOT NULL, BUDGET_VERSION_ID NUMBER(15), LAST_UPDATE_DATE DATE NOT NULL, . . . . . . ) PARTITION BY RANGE (PERIOD_NAME) ( PARTITION jan04_per VALUES LESS THAN (JAN-2005), PARTITION feb04_per VALUES LESS THAN (FEB-2005) . . . . . . );

Partitioning Methods- List


Partitions are mapped to an enumerated list. Allows data to be mapped to partitions by specifying a list of discrete values for the partitioning key. List partitioning does not support multi-column partition keys. List partitioning was introduced in Oracle9i.

Partitioning Methods - List


 Partition the OM runtime tables by open_flag
OE_ORDER_LINES_ALL (44,519,880 rows)

Open Orders (192,403 rows)

Closed Orders (40,169,899 rows)

Partitioning Methods
List Partitioning Example:
CREATE TABLE OE_ORDER_LINES_ALL (LINE_ID NUMBER NOT NULL, ORG_ID NUMBER, HEADER_ID NUMBER NOT NULL, LINE_TYPE_ID NUMBER NOT NULL, LINE_NUMBER NOT NULL NUMBER ORDERED_ITEM VARCHAR2(2000), OPEN_FLAG VARCHAR2(1) NOT NULL, . . . . . . ) PARTITION BY LIST (open_flag) ( PARTITION open_par VALUES (Y), PARTITION close_par VALUES (N) );

Partitioning Methods - Hash


 Applies a hash function to the partitioning key to stripe data into partitions.  Controls the physical placement of data across a fixed number of partitions.  Hash partitioning was introduced in Oracle8i.  Generally useful for Applications batch programs which use parallel workers for which block contention is significant.
 Order Import  Invoice Import

Partitioning Methods - Hash


 Partition the OM runtime tables by the primary key
OE_ORDER_LINES_ALL (44,519,880 rows)

Hash1 5,045,430 rows

Hash2 5,045,402 rows

Hash3 5,043,712 rows

Hash4 5,042,929 rows

Hash5 5,046,608 rows

Hash6 5,045,024 rows

Hash7 5.048.923 rows

Hash8 5,044,312 rows

Partitioning Methods
Hash Partitioning Example:
CREATE TABLE OE_ORDER_LINES_ALL (LINE_ID NUMBER NOT NULL, ORG_ID NUMBER, HEADER_ID NUMBER NOT NULL, LINE_TYPE_ID NUMBER NOT NULL, LINE_NUMBER NOT NULL NUMBER ORDERED_ITEM VARCHAR2(2000), OPEN_FLAG VARCHAR2(1) NOT NULL, . . . . . . ) PARTITION BY HASH (LINE_ID) PARTITIONS 8 . . . . . .;

Partitioning Methods - Composite


 Data is partitioned by using both the range method and then sub-partitioned by the hash method.  Composite partitioning combines the best of both worlds (range and hash) by allowing logical groupings at the partition level and handling data skew within the sub-partitions.  Composite partitioning was introduced in Oracle8i.

Partitioning Methods Composite


 The Workflow Runtime tables use composite partitioning.
WF_ITEM_ACTIVITY_STATUSES (82,585,860 rows)

Purchase Order Item Type (13,319,867 rows)

Order Item Type (3,402,130 rows)

Item Key Hash


hash1 hash2 hash3 hash4 hash5 hash6 hash7 hash8 hash1 hash2 hash3 hash4 hash5 hash6 hash7 hash8

Partitioning Methods
Composite Partitioning Example:
create table wf_item_activity_statuses partition by range (item_type) subpartition by hash (item_key) subpartitions 8 (partition wf_item1 values less than ('A1'), partition wf_item2 values less than ('AM'), partition wf_item3 values less than ('AP'), partition wf_item4 values less than ('AR'), partition wf_item5 values less than ('AZ'), . . . partition wf_item48 values less than ('OE'), partition wf_item49 values less than ('OF'), partition wf_item50 values less than ('OK'), partition wf_item51 values less than ('OL'), . . . partition wf_item56 values less than ('PO'), partition wf_item57 values less than ('PQ'), partition wf_item58 values less than ('PR'), partition wf_item59 values less than ('QA'), . . . );

<Insert Picture Here>

Index Partitioning

Index Partitioning
Indexes can be partitioned like tables Partitioned or nonpartioned indexes can be used with partitioned or nonpartitioned tables Partioned indexes can be
Global or local Prefixed or nonprefixed
Table Index Table Index

Normal

Partitioned

Index Partitioning
Index Partitioning Methods
Global Non-partitioned (prefixed) Global Partitioned (prefixed) Local Partitioned Prefixed Local Partitioned Non-Prefix

Global & Local Index Partitioning


Global nonpartitioned index Index Partitioning Methods Global non-partitioned index Global partitioned prefixed index Local prefixed index Local non-prefixed index

Global partitioned Index

Table partitions Local prefixed index Local nonprefixed index

Global Indexes
Index Partitioning Methods
Global Non-partitioned (prefixed) Use Global Non-partitioned indexes for all indexes which are not prefixed by the table partition key. Global Range Partitioned (prefixed) Generally not useful for Applications. Global Hash Partitioned (prefixed) new in 10g Extremely useful for right growing indexes experiencing contention due to high levels of concurrency. Allows the index to be partitioned without affecting the table.

Local Indexes
Index Partitioning Methods
Local Partitioned Prefixed Use in place of indexes which already contain the partition key as a prefix. Local Partitioned Non-Prefix Generally not useful for Applications. Should only be used when all the queries using the local (non-prefixed) index always include the partition key filter.

Global Index Partitioning


Index Partitioning Example
GL_BALANCES table is partitioned by range on the period_name

Existing Indexes New Index Name: Columns: Index Type: ======================================================================= GL_BALANCES_N1 (CODE_COMBINATION_ID,PERIOD_NAME) Global GL_BALANCES_N2 (PERIOD_NAME) Not Needed GL_BALANCES_N3 (PERIOD_NUM, PERIOD_YEAR) Global GL_BALANCES_N4 (TEMPLATE_ID) Global

Local & Global Indexes


Index Partitioning Example
WF_ITEM_ACTIVITY_STATUSES table is partitioned using composite partitioning (range on item_type, sub-partitioned by hash on the item_key).
Existing Indexes New Index Name: Columns: Index Type: ============================================================================= WF_ITEM_ACTIVITY_STATUSES_N1 (ACTIVITY_STATUS,ITEM_TYPE) Global WF_ITEM_ACTIVITY_STATUSES_N2 (NOTIFICATION_ID) Global WF_ITEM_ACTIVITY_STATUSES_N3 (DUE_DATE, ITEM_TYPE) Global WF_ITEM_ACTIVITY_STATUSES_N4 (ASSIGNED_USER, ITEM_TYPE) Global WF_ITEM_ACTIVITY_STATUSES_PK (ITEM_TYPE,ITEM_KEY,PROCESS_ACTIVITY) Local

Useful Data Dictionary Views


Table Partitioning

View Name DBA_TABLES DBA_PART_TABLES DBA_TAB_PARTITIONS DBA_PART_KEY_COLUMNS DBA_PART_HISTOGRAMS

Purpose
Table structure, Partition Y/N Partition type, default values Partitions detail Partition keys Histogram Statistics

L T T P P P P

DBA_PART_COL_STATISTICS Column and Histogram Stats. T = per Table P = per Partition

Useful Data Dictionary Views


Index Partitioning

Name DBA_SUBPART_KEY_COLUMNS DBA_IND_PARTITIONS DBA_IND_SUBPARTITIONS DBA_PART_INDEXES

Purpose Sub-partition key Information Index Partition details Index Sub-partition details Partitioned Index Definition

L T P P I

T = per Table I = per index P = per Partition

<Insert Picture Here>

Customer Partitioning Case Studies

Partitioning & Table Compression


Example OE_ORDER_LINES_ALL
create table oe_order_lines_all_comp parallel compress tablespace A_TXN_DATA_01 nologging;

Before table compression


select segment_name, blocks from dba_segments where segment_name = 'OE_ORDER_LINES_ALL'; SEGMENT_NAME OE_ORDER_LINES_ALL BLOCKS 4322352 ---------------------------------------------------- -------

After table compression


select segment_name, blocks from dba_segments where segment_name = 'OE_ORDER_LINES_ALL_COMP'; SEGMENT_NAME OE_ORDER_LINES_ALL_COMP BLOCKS 1234957 (~3X) ---------------------------------------------------- -------

Case Study 1:
General Ledger (GL) Partitioning
Customer Case Study
International Bank Data Volumes GL_JE_LINES (1.1 Billion rows) GL_CODE_COMBINATIONS (203 Million rows) GL_BALANCES (1.3 Billion Rows) Partitioning Method: Range Partition Key: set_of_books_id # of Partitions: 34 Achieved Throughput: 11.4 million journal lines imported and posted per hour

Case Study 2:
General Ledger (GL) Partitioning
Customer Case Study
Australian Bank Data Volumes GL_JE_LINES (650 million rows) GL_CODE_COMBINATIONS (8.5 Million rows) GL_BALANCES (200 million Rows) Partitioning Method: Range Partition Key: period_name # of Partitions: 109 Achieved Throughput: 7.5 million journal lines imported and posted per hour

Case Study 3:
Accounts Payable (AP) Partitioning Partitioning Case Study
Payables Trial Balances Data Volumes AP_LIABILITY_BALANCE (70,567,250 rows) Partitioning Method: Hash Partition Key: ORG_ID # of Partitions: 32 Achieved Throughput: Trial Balance report runtime reduced from 2 hours to 10 minutes.

<Insert Picture Here>

Information Lifecycle Management (ILM)

The Data Lifecycle


Is all this data needed or required to be available online?
Will require more storage and memory Increased data will impact performance for both online and batch processing

Go-live

This Month

This Year

Previous Years

Active

Less Active

Historical

Archive

Tables continue to grow Data comes from many sources such as Transactional (Fiscal/Non-Fiscal) (e.g. General Ledger) Transient data (i.e. login, concurrent requests etc..) Reference/seed data

What is ILM?
Information Lifecycle Management encompasses the following:
Policies which define how to manage the data Processes which actually manage the data Including Archiving & Purging and Partitioning Software which performs the policies & processes RDBMS Partitioning functionality (ILM Advisor) Purge & Archive concurrent programs Hardware where the data is stored

Partitioning, Purging and Archiving are all integral parts of the ILM lifecycle

Implementing ILM

1. Define Data Classes 3. Manage Access and Migration of Data by Class


Active Less Active Historical

2. Create Storage Tiers for the Data Classes

High Performance Low Cost Storage Tier Storage Tier

Historical Storage Tier

D E M O N S T R A T I O N

ILM Assistant

<Insert Picture Here>

Archiving & Purging in the E-Business Suite

Purging & Archiving


Archive Periodically saving data (with reduced attributes) e.g. transaction data
Periodically After Archive

Purge Deleting data that you no longer need e.g. transient data or useless data such as dead leads or closed orders.

Statutory Requirements

Space

Historical Requirements

Performance

Improves Upgrade performance (Less Data.. Less Work)

Purging & Archiving


Fiscal Data - Financial Data that can be used to represent a profit or a loss e.g. GL_JOURNALS Typically, you should purge and archive non-fiscal data such as sales leads, sales opportunities

Purging & Archiving


Utilize the standard Oracle supplied programs
Purge/Archive programs provided at the module level in the E-Business Suite Many documented standard programs available ~214 purge programs in 11.5.10 ~260 purge programs in R12 Financials
GL, AR, AP, Costing, Cash Management, HR, FA, PA

Manufacturing
Order Management, INV, BOM, WIP, QP, Shipping, Purchasing

Purge and Archive Programs


Product AOL General Ledger Assets Payables Receivables Order Management Purchasing Concurrent Program Name Purge Concurrent Request /Manager Data Purge "Sign on" Audit data GL Archive and Purge Mass Additions Purge Report Purge accounting tables Payables Purge Archive and Purge Order Purge Purge and Archive POs

Customer Example:
Typical Purge Programs run
Concurrent Program Name
Purge Debug Log Purge Cost Information Purge Signon Audit data Payables Open Interface Purge Purge Invoice Extract Output Table Catalog Data Purge Purge Concurrent Request/Manager Data Purge System Saved Requisition Purge Interface Tables Purge Obsolete Workflow Runtime Data BEE Batch Process (Purge) Delete/Purge Timecards (OTL)

Concurrent Program
FNDLGPRG CSTCSPCT FNDSCPRG APXIIPRG RAXINVPG POXCDXPG FNDCPPUR POXSSPG PNVPURGE FNDWFPR PAYLINK(PURGE) DELPURTC

<Insert Picture Here>

Demonstration Purge Portal

Navigation
Accessing the Purge Portal
System Administrator > Oracle Applications Manager >Purging/Critical Activities

Purge Portal Features


Purge Portal introduced in 11i10
Single purge/archive management console Purge programs can be configured, initiated and monitored Set the execution frequency as well view history of purge programs.

Further enhanced in Release 12


Purge programs tagged with the purge concurrent program type.

Purging Purge Portal

More Purge Programs in Release 12


Service Request Purge Faster WF Purge
Rewritten to use bulk processing

Email Fulfillment Purge Export and Purge All GL Interface Data Approvals Management Transaction Data Purge Purge Secured Payment Data Service Contracts Purge Credit Card Shipping Purge Task Purge Program Web Analytics: Purge Tracking Data Many others

Purging & Archiving


Oracle is committed to providing Purge/Archive solutions for the key transactional entities Oracle continues to enhance its purge and archive offering The native E-Business Suite purge and archive programs are synchronized with the data model
Avoids having to re-apply the same patch to multiple schemas when using non-native purge/archive solutions.

Further Information
MetaLink & Product User Guides Please refer to MetaLink (examples)
138264.1 General Ledger Archive/Purge FAQ 144431.1 Fixed Assets Archive/Purge FAQ 136919.1 General Ledger Archive/Purge Setup and Usage

AR 11.5.10 User Guide Chapter:1 Archive & Purge

GL 11.5.10 User Guide Chapter: 8 Maintenance

AP 11.5.10 User Guide Chapter: 10 Resource Management

<Insert Picture Here>

Summary

Options Available

ILM Archive
Data Storage

Purge Partitioning
Data Access

Partitioning in Oracle Applications


What you need to consider
Functional use of the table (i.e. what exactly does the table store?). Which modules or business flows use this table? Growth rates and access patterns of this table. Review the Statspack or AWR Repository for the relevant SQL statements. Determine the optimal table partitioning method based on your implementation and based on the data points above. Evaluate different indexing methods, global vs. local partitioned indexes. Leverage the ILM Assistant to implement partitioning.

Implementing Purging & Archiving


Consider
Data Retention Requirements Recovery Requirements Data access methods and security Storage topology and costs

Define
Your archive and purge policy

We would like to hear from you


<Insert Picture Here>

If you have any questions or experiences you would like to share regarding partitioning and/or purging and archiving in the E-Business Suite, please send an e-mail: mohsin.sameen@oracle.com

MetaLink References

Note #:

Description:

138264.1 144431.1 136919.1 248857.1

General Ledger Archive and Purge - Frequently Asked Questions Oracle Assets Archive, Purge and Restore Processes General Ledger Archive/Purge Setup and Usage Oracle Applications Tablespace Model Release 11i - Tablespace Migration Utility

<Insert Picture Here>

Questions & Answers

Potrebbero piacerti anche