SlideShare a Scribd company logo
2
Most read
3
Most read
4
Most read
© Copyright IBM Corporation 2009
IBM Software Group
IMS Buffer Pool Tuning
Manoj Kaveri
Cert IMS DBA
Large Healthcare Payor Account in US
Oct 2010
IMS Storage
 VSAM Control Interval : VSAM works with a logical data area known as a
control interval (CI). The default CI size is 4K bytes, but it can be up to
32K bytes. The CI contains data records, unused space, and a CI
descriptor field.
 Most commonly used VSAM file in IMS.
- Key Sequence Data Set (KSDS)
Contain two components: INDEX and DATA
-Entry Sequence Data Set (ESDS)
Contain single component: DATA.
 OSAM Block Size : OSAM works with a logical data area know as a
block size.
 VSAM dataset can store data up to 4GB & OSAM up to 8GB
 Identify the file type in DBD
DBD NAME=XXLA01DB,
ACCESS=(HIDAM,VSAM) or ACCESS=(HIDAM,OSAM)
AMS Share-Net | Oct 2010
Innovative Improved
Usability
Open
Standards
IMS Database Types
 Full- Function Database
HSAM,HISAM,HDAM,HIDAM
 Fast-Path Database
DEDB,MSDB
 HALDB Database
PHDAM,PHIDAM,PSINDEX
 Identify the database type based on access methods in DBD
DBD NAME=XXLA01DB,
ACCESS=(HIDAM,VSAM)
Innovative Improved
Usability
Open
Standards
4
Introduction
 Buffer poolBuffer pool are areas of virtual storage that temporarily stores data
during run-time.
 Allocate sufficient buffers in the IMS database buffer pools to prevent
I/O that results from an application program having to reread data
previously brought into the pool.
 Careful choice of sub pool sizes and matching against database block
sizes and frequency of database references is required.
 Type of buffer pools – Based on file type
1. VSAM Buffer Pool
- Define VSAM sub pools using the VSRBF control statement
2. OSAM Buffer Pool
- Define OSAM subpools are defined using the IOBF.
 Buffer Sizes can be 5 KB, 1 KB, 2 KB, 4 KB, 8 KB and multiple of 4KB.
 Buffer Size should be selected based on CI SIZE of VSAM or BLOCK
SIZE of OSAM file.
Innovative Improved
Usability
Open
Standards
AMS Share-Net | Oct 2010
5
Buffer Pool Objective
Eliminating I/O reduces
CPU costs
Increase Number of
Page Hit Ratio.
Reduce CPU Time
Reduce Number of
Page SWAP.
I/O Avoidance
-Keep data in
memory
-Buffer Pool.
Optimize memory Usage
Batch Jobs
-Achieve your batch
requirements
- Maximize throughput
Transactions
-Improve productivity
-Improve clients
satisfaction
Reduce Elapse Time
AMS Share-Net | Oct 2010
6
Buffer Pool Component Located
DLI NOT WAIT
IMS
Controller
IMS DB
IMS
PC
PROGRAM ELAPSE TIME
IMS Buffer Space
IMS DB
IMS Program
DLI ELAPSE TIME APPLICATION
DLI IWAIT
DB Access
AMS Share-Net | Oct 2010
7
How to meet the performance Objective ?
Business Scenario :

Buffer spool option be selected only if the user has established
performance objectives and know how the current system is
performing.
 Understand Current Business Scenario :
• Batch job/Online Program is access Large number of IMS Database.
• Database type and Access methods.
-Database type : Full-Function, Fast-Path, HALDB
-Access Method : HISAM,HDAM,HIDAM,PHDAM,PHIDAM
• IMS Storage and Control Interval Verification.
- Verify CI Size of VSAM INDEX & DATA Component.
- Verify Block Size of OSAM file.
AMS Share-Net | Oct 2010
8
How to meet the performance Objective ?
Problem Statement :
 Buffer pool was not defined according to the VSAM CI or OSAM Block
size.
 Buffer pool was not been divide into sub-pool for Data (D –VSAM Data
component, ESDS) and Index (I- VSAM Index Component).
 Common buffer control card has been used across all the batch jobs.
 Existing production code :
DFSVSAMP DD DSN=TEST.PROCLIB (XXVSAMP)
18432,4
16384,4
12288,200
8192,700
IOBF=(2048,4,N,N)
IOBF=(4096,500,N,N)
IOBF=(6144,300,N,N)
AMS Share-Net | Oct 2010
9
How to meet the performance Objective ?
Solution:
 IMS may build multiple VSAM pools. Each pool may have multiple sub pools.
 A VSAM pool is defined by a POOLID= statement in the DFSVSMxx member for
online systems or the DFSVSAMP data set for batch and utility jobs.
 Suggest Changes :
POOLID=SP01
VSRBF=8192, 700, D
DBD= XTDBD01 (1, SP01)
POOLID=SP02
VSRBF=12288, 200, D
VSRBF=10240, 500, I
DBD= XTDBD02 (1, SP02)
IOBF=(2048,4,Y,Y)
OPTIONS, VSAMFIX= (BFR, IOB), BGWRT= (YES, 50)
By dividing buffer pools in data and index component number of page hit ratio
can be increased and reduce the I/O transfer rate/sec, which results in reduction
of elapse time, CPU units, I/O transfer counts.
 Buffer pool has been defined according to the VSAM CI or OSAM Block Size.
AMS Share-Net | Oct 2010
10
How to meet the performance Objective ?
Change Desc:Change Desc:
 VSAM ESDS Buffer poolVSAM ESDS Buffer pool
POOLID=SP01 /* POOLID: Define buffer pool name, SPO1: is buffer pool name. Its user define namePOOLID=SP01 /* POOLID: Define buffer pool name, SPO1: is buffer pool name. Its user define name
VSRBF=8192, 700, D /*-VSRBF: Sub pools in the pool are defined by the VSRBF= statements which followVSRBF=8192, 700, D /*-VSRBF: Sub pools in the pool are defined by the VSRBF= statements which follow
/*the POOLID= statement - 8192-: Buffer Size -Sub pool has a buffer size of 8192,/*the POOLID= statement - 8192-: Buffer Size -Sub pool has a buffer size of 8192,
/*700 : Number of buffer pool/No. of block in SPO1 buffer pool./*700 : Number of buffer pool/No. of block in SPO1 buffer pool.
D : Data component will use this buffer pool SP01.D : Data component will use this buffer pool SP01.
 VSAM KSDS Buffer poolVSAM KSDS Buffer pool
POOLID=SP02POOLID=SP02
VSRBF=12288,200,D / * D : Indicates that data component can use this subpoolVSRBF=12288,200,D / * D : Indicates that data component can use this subpool
VSRBF=10240,500,I /* I : Indicated the only index component can use this subpool.VSRBF=10240,500,I /* I : Indicated the only index component can use this subpool.
DBD=XTDBD02 (1,SP02) /* XTDBD02 : DBD name corresponding to dataset.DBD=XTDBD02 (1,SP02) /* XTDBD02 : DBD name corresponding to dataset.
Job1
Shared
Buffer Pools
Job2
Online Scr
Remember:Remember:
Performance dependencePerformance dependence
on whether shared bufferon whether shared buffer
used by multiple programused by multiple program
during run-timeduring run-time
AMS Share-Net | Oct 2010
11
Test Results..
 B&UT Testing : IBM GR Team – Test Results
Description Base Converted
Run1
CPU UNITS 3,026,234 2,843,993
CPU Time 00:01:14.88 00:01:02.37
SERV UNITS 3,092,247 2,906,908
SRB UNITS 51,595 48,524
I/O UNITS 14,418 14,391
Execution Time
(step XNC100X)
00:10:55.09 00:05:30.62
DLI Statistics Report
Calls Base Convert
GU 47 41
GN 83,342 83,343
GNP 2,062,941 2,062,981
GHU 6 6
DB
(TOTAL) 2,146,342 2,146,383
CHKP 42 42
LOG 294 294
Remember 1:Remember 1:
Performance should bePerformance should be
compared based on No. ofcompared based on No. of
DLI calls issued & No. ofDLI calls issued & No. of
record accessed inrecord accessed in
program.program.
AMS Share-Net | Oct 2010
12
Test Results..
 SIT Testing : Client Technical Expert Group – Test Results
Description Base Converted
Run1
Converted Run2 Converted Run3
CPU UNITS 10,264,284 9,914,169 9,538,354 9,031,724
CPU Time 00:04:06.28 00:04:05.32 00:03:56.02 00:03:36.71
SERV UNITS 10,584,075 10,227,066 9,842,507 9,325,137
SRB UNITS 233,312 230,039 226,202 217,536
I/O UNITS 86,479 82,858 77,951 75,877
Execution Time (step
XNC100X)
00:30:06.43 00:24:18.70 00:20:10.64 00:12:16.73
Remember 2:Remember 2:
Whether the IMS1/IMS2Whether the IMS1/IMS2
region database has beenregion database has been
re-org recentlyre-org recently
AMS Share-Net | Oct 2010
13
Business Benefits…
 Buffer pool concept will be applicable to batch & online application.
 By proposing IMS buffer-pool concept we are able to reduce 10-20% of elapse
time, CPU units, I/O transfer counts.
 Implementing buffer-pool concept into there respective account will add values to
clients and will be able to meet SLA, as it will reduce application elapse time.
 Dividing buffer pools in data and index component number of page hit ratio can be
increased and reduce the I/O transfer rate/sec
 Supports additional new opts Included – BGWRT (Back Ground Write) % used for
parallel write operation to database and buffer pool can be released during
database processing.
IMS
BufferIMS
IMS
AMS Share-Net | Oct 2010
14
Performance Dependent Parameters…
 Performance of XNC100JD dependent on the below parameters
 Region Parameters:
• Whether these database are used by other program during the batch job
execution.
• Shared buffer used by other program for loading the data.
• Whether the IMS1/IMS2 region database has been reorganized recently.
 Application Parameters:
• No. of record retrieved from the database..
• No. of DLI call issued by the program.
• BGWRT- The best percentage value to be specified will change depending on
whether the program is re- accessing the buffers or not.
• Background Write: In general, maximum benefit occurs in sequential update
processing.
AMS Share-Net | Oct 2010
15
Implementation Method…
 Batch Jobs
- Update the DFSVSAMP DD Control Cards
 Online Program
- DFSVSMxx member in IMS.PROCLIB for online systems
Example 1:
POOLID=VSM1
VSRBF=4096,3000,D
VSRBF=12288,2000,I
VSRBF=32768,12,D
Since the second VSRBF statement builds an index subpool, the first and third
subpools may only be used for data components. The second subpool may only
be used for index components.
Assumption:Assumption:
Application is IMS BatchApplication is IMS Batch
or Online Programor Online Program
AMS Share-Net | Oct 2010
16
For more information
 IMS Full Function Database Use of VSAM Buffer Pools
-http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/FLASH10240
 IMS Buffer Pool Analyzer
- http://publib.boulder.ibm.com/infocenter/dzichelp/v2r2/index.jsp?
topic=/com.ibm.imstools.bpl.doc.ug/bplucon_hilites.htm
 IMS Performance Monitoring and Tuning Update
-ibm.com/redbooks
AMS Share-Net | Oct 2010
17
IMS Buffer Pool is …
… Just
Great !
THANK YOU …..ANY QUESTIONS
AMS Share-Net | Oct 2010

More Related Content

What's hot (20)

PDF
Practical Partitioning in Production with Postgres
EDB
 
PPT
DB2 and storage management
Craig Mullins
 
PDF
Ibm db2 interview questions and answers
Sweta Singh
 
PDF
DB2 for z/OS Bufferpool Tuning win by Divide and Conquer or Lose by Multiply ...
John Campbell
 
PPTX
Ibm db2
aditi212
 
PDF
zOSMF SDSF_ShareLab_V2R5.pdf
Marna Walle
 
PPT
DB2 Interview Questions - Part 1
ReKruiTIn.com
 
PDF
IBM DB2 for z/OS Administration Basics
IBM
 
PDF
DB2 LUW - Backup and Recovery
imranasayed
 
DOC
Jcl interview questions
ganjigirish
 
PPT
Introduction to sql
VARSHAKUMARI49
 
PDF
Tanel Poder - Performance stories from Exadata Migrations
Tanel Poder
 
PDF
100 COOL MAINFRAME TIPS
Nirmal Pati
 
DOC
Mainframe interview
Shivankoo Sharma
 
PPSX
Introducing the eDB360 Tool
Carlos Sierra
 
PDF
IMS DC Self Study Complete Tutorial
Srinimf-Slides
 
PPT
Cics testing and debugging-session 7
Srinimf-Slides
 
PDF
DB2 TABLESPACES
Rahul Anand
 
PDF
ClickHouse Deep Dive, by Aleksei Milovidov
Altinity Ltd
 
DOC
Top jcl interview questions and answers job interview tips
jcltutorial
 
Practical Partitioning in Production with Postgres
EDB
 
DB2 and storage management
Craig Mullins
 
Ibm db2 interview questions and answers
Sweta Singh
 
DB2 for z/OS Bufferpool Tuning win by Divide and Conquer or Lose by Multiply ...
John Campbell
 
Ibm db2
aditi212
 
zOSMF SDSF_ShareLab_V2R5.pdf
Marna Walle
 
DB2 Interview Questions - Part 1
ReKruiTIn.com
 
IBM DB2 for z/OS Administration Basics
IBM
 
DB2 LUW - Backup and Recovery
imranasayed
 
Jcl interview questions
ganjigirish
 
Introduction to sql
VARSHAKUMARI49
 
Tanel Poder - Performance stories from Exadata Migrations
Tanel Poder
 
100 COOL MAINFRAME TIPS
Nirmal Pati
 
Mainframe interview
Shivankoo Sharma
 
Introducing the eDB360 Tool
Carlos Sierra
 
IMS DC Self Study Complete Tutorial
Srinimf-Slides
 
Cics testing and debugging-session 7
Srinimf-Slides
 
DB2 TABLESPACES
Rahul Anand
 
ClickHouse Deep Dive, by Aleksei Milovidov
Altinity Ltd
 
Top jcl interview questions and answers job interview tips
jcltutorial
 

Similar to IMSBufferpool Tuning concept AMS presentation v01 (20)

PDF
optimizing_ceph_flash
Vijayendra Shamanna
 
PDF
IBM DB2 Analytics Accelerator Trends & Directions by Namik Hrle
Surekha Parekh
 
PDF
IBM Analytics Accelerator Trends & Directions Namk Hrle
Surekha Parekh
 
PPTX
Db2 analytics accelerator on ibm integrated analytics system technical over...
Daniel Martin
 
PDF
Ims05 ims 100 k benchmark
Robert Hain
 
PDF
SMP / Extended orientation for IBM mainframes
tejaswinireddy524612
 
PDF
Big Lab Problems Solved with Spectrum Scale: Innovations for the Coral Program
inside-BigData.com
 
PPT
Ibm ims tools san ramon oct 2011
evgeni77
 
PPTX
Ceph Day Taipei - Accelerate Ceph via SPDK
Ceph Community
 
PDF
Webcenter application performance tuning guide
Vinay Kumar
 
PDF
WebSphere Technical University: Top WebSphere Problem Determination Features
Chris Bailey
 
PDF
System z Technology Summit Streamlining Utilities
Surekha Parekh
 
PDF
AdminOps-Part-II-Installation.pdf of okera
bogonias1
 
PPTX
Using SAS GRID v 9 with Isilon F810
Boni Bruno
 
PPTX
Ceph Day Berlin: Ceph on All Flash Storage - Breaking Performance Barriers
Ceph Community
 
PDF
DB2 Design for High Availability and Scalability
Surekha Parekh
 
ODP
IMS v12 performance
evgeni77
 
PDF
Ims12 workbench data visualization - IMS UG May 2014 Sydney & Melbourne
Robert Hain
 
PPT
DB2 for z/O S Data Sharing
Surekha Parekh
 
PDF
Spectrum Scale final
Joe Krotz
 
optimizing_ceph_flash
Vijayendra Shamanna
 
IBM DB2 Analytics Accelerator Trends & Directions by Namik Hrle
Surekha Parekh
 
IBM Analytics Accelerator Trends & Directions Namk Hrle
Surekha Parekh
 
Db2 analytics accelerator on ibm integrated analytics system technical over...
Daniel Martin
 
Ims05 ims 100 k benchmark
Robert Hain
 
SMP / Extended orientation for IBM mainframes
tejaswinireddy524612
 
Big Lab Problems Solved with Spectrum Scale: Innovations for the Coral Program
inside-BigData.com
 
Ibm ims tools san ramon oct 2011
evgeni77
 
Ceph Day Taipei - Accelerate Ceph via SPDK
Ceph Community
 
Webcenter application performance tuning guide
Vinay Kumar
 
WebSphere Technical University: Top WebSphere Problem Determination Features
Chris Bailey
 
System z Technology Summit Streamlining Utilities
Surekha Parekh
 
AdminOps-Part-II-Installation.pdf of okera
bogonias1
 
Using SAS GRID v 9 with Isilon F810
Boni Bruno
 
Ceph Day Berlin: Ceph on All Flash Storage - Breaking Performance Barriers
Ceph Community
 
DB2 Design for High Availability and Scalability
Surekha Parekh
 
IMS v12 performance
evgeni77
 
Ims12 workbench data visualization - IMS UG May 2014 Sydney & Melbourne
Robert Hain
 
DB2 for z/O S Data Sharing
Surekha Parekh
 
Spectrum Scale final
Joe Krotz
 
Ad

Recently uploaded (16)

PPTX
MY lovely friend in my life sweet and young lovely baudi
TirthankarBose9
 
PDF
The-Power-of-Public-Speaking- communictaion skills.pdf
holin7
 
PPTX
Top 5 Useful Apps Every Parent Should Know
Michael Carter
 
PPTX
At the End of the Tree_21122006_21122006
dhruvchudasama212006
 
PPTX
The Journey of Life Full Presentation.pptx
AlyssaKaytePalgan
 
PPTX
Quetta_Tag_Lifestyle balochistan qta.pptx
MuhammadAkramKhan9
 
PPTX
Lesson4_If I Could Talk To Animals - Public speaking
NhungNgLHoi
 
PPTX
Vice-and-Drug-Education-and-Control-Lesson-1-4 (1).pptx
heycablao
 
PPTX
D o g s l i f e s t y l e _ Pres entation.pptx
TarekMneimneh4
 
PPTX
Eating Habit Morning meals and your blood glucose.pptx
RIVALDOMALAWAT1
 
PPTX
Faisal_Town_Quetta_Presentation ccc.pptx
MuhammadAkramKhan9
 
PPTX
Difference of Karate and KungFu (martial art)
winsoncheok1
 
PDF
Astrological Marriage Guide: Tamil Jathagam Predictions Explained
One Page Horoscope
 
PPTX
Presentation 3.pdf_20250724_202404_0000.pptx
AttaullahKhan837818
 
PPTX
C a t s l i f e s t y l e_Presentation.pptx
TarekMneimneh4
 
PDF
Pert Smart Home Automation --- Catalogue
Pert Infoconsulting Private Limited
 
MY lovely friend in my life sweet and young lovely baudi
TirthankarBose9
 
The-Power-of-Public-Speaking- communictaion skills.pdf
holin7
 
Top 5 Useful Apps Every Parent Should Know
Michael Carter
 
At the End of the Tree_21122006_21122006
dhruvchudasama212006
 
The Journey of Life Full Presentation.pptx
AlyssaKaytePalgan
 
Quetta_Tag_Lifestyle balochistan qta.pptx
MuhammadAkramKhan9
 
Lesson4_If I Could Talk To Animals - Public speaking
NhungNgLHoi
 
Vice-and-Drug-Education-and-Control-Lesson-1-4 (1).pptx
heycablao
 
D o g s l i f e s t y l e _ Pres entation.pptx
TarekMneimneh4
 
Eating Habit Morning meals and your blood glucose.pptx
RIVALDOMALAWAT1
 
Faisal_Town_Quetta_Presentation ccc.pptx
MuhammadAkramKhan9
 
Difference of Karate and KungFu (martial art)
winsoncheok1
 
Astrological Marriage Guide: Tamil Jathagam Predictions Explained
One Page Horoscope
 
Presentation 3.pdf_20250724_202404_0000.pptx
AttaullahKhan837818
 
C a t s l i f e s t y l e_Presentation.pptx
TarekMneimneh4
 
Pert Smart Home Automation --- Catalogue
Pert Infoconsulting Private Limited
 
Ad

IMSBufferpool Tuning concept AMS presentation v01

  • 1. © Copyright IBM Corporation 2009 IBM Software Group IMS Buffer Pool Tuning Manoj Kaveri Cert IMS DBA Large Healthcare Payor Account in US Oct 2010
  • 2. IMS Storage  VSAM Control Interval : VSAM works with a logical data area known as a control interval (CI). The default CI size is 4K bytes, but it can be up to 32K bytes. The CI contains data records, unused space, and a CI descriptor field.  Most commonly used VSAM file in IMS. - Key Sequence Data Set (KSDS) Contain two components: INDEX and DATA -Entry Sequence Data Set (ESDS) Contain single component: DATA.  OSAM Block Size : OSAM works with a logical data area know as a block size.  VSAM dataset can store data up to 4GB & OSAM up to 8GB  Identify the file type in DBD DBD NAME=XXLA01DB, ACCESS=(HIDAM,VSAM) or ACCESS=(HIDAM,OSAM) AMS Share-Net | Oct 2010 Innovative Improved Usability Open Standards
  • 3. IMS Database Types  Full- Function Database HSAM,HISAM,HDAM,HIDAM  Fast-Path Database DEDB,MSDB  HALDB Database PHDAM,PHIDAM,PSINDEX  Identify the database type based on access methods in DBD DBD NAME=XXLA01DB, ACCESS=(HIDAM,VSAM) Innovative Improved Usability Open Standards
  • 4. 4 Introduction  Buffer poolBuffer pool are areas of virtual storage that temporarily stores data during run-time.  Allocate sufficient buffers in the IMS database buffer pools to prevent I/O that results from an application program having to reread data previously brought into the pool.  Careful choice of sub pool sizes and matching against database block sizes and frequency of database references is required.  Type of buffer pools – Based on file type 1. VSAM Buffer Pool - Define VSAM sub pools using the VSRBF control statement 2. OSAM Buffer Pool - Define OSAM subpools are defined using the IOBF.  Buffer Sizes can be 5 KB, 1 KB, 2 KB, 4 KB, 8 KB and multiple of 4KB.  Buffer Size should be selected based on CI SIZE of VSAM or BLOCK SIZE of OSAM file. Innovative Improved Usability Open Standards AMS Share-Net | Oct 2010
  • 5. 5 Buffer Pool Objective Eliminating I/O reduces CPU costs Increase Number of Page Hit Ratio. Reduce CPU Time Reduce Number of Page SWAP. I/O Avoidance -Keep data in memory -Buffer Pool. Optimize memory Usage Batch Jobs -Achieve your batch requirements - Maximize throughput Transactions -Improve productivity -Improve clients satisfaction Reduce Elapse Time AMS Share-Net | Oct 2010
  • 6. 6 Buffer Pool Component Located DLI NOT WAIT IMS Controller IMS DB IMS PC PROGRAM ELAPSE TIME IMS Buffer Space IMS DB IMS Program DLI ELAPSE TIME APPLICATION DLI IWAIT DB Access AMS Share-Net | Oct 2010
  • 7. 7 How to meet the performance Objective ? Business Scenario :  Buffer spool option be selected only if the user has established performance objectives and know how the current system is performing.  Understand Current Business Scenario : • Batch job/Online Program is access Large number of IMS Database. • Database type and Access methods. -Database type : Full-Function, Fast-Path, HALDB -Access Method : HISAM,HDAM,HIDAM,PHDAM,PHIDAM • IMS Storage and Control Interval Verification. - Verify CI Size of VSAM INDEX & DATA Component. - Verify Block Size of OSAM file. AMS Share-Net | Oct 2010
  • 8. 8 How to meet the performance Objective ? Problem Statement :  Buffer pool was not defined according to the VSAM CI or OSAM Block size.  Buffer pool was not been divide into sub-pool for Data (D –VSAM Data component, ESDS) and Index (I- VSAM Index Component).  Common buffer control card has been used across all the batch jobs.  Existing production code : DFSVSAMP DD DSN=TEST.PROCLIB (XXVSAMP) 18432,4 16384,4 12288,200 8192,700 IOBF=(2048,4,N,N) IOBF=(4096,500,N,N) IOBF=(6144,300,N,N) AMS Share-Net | Oct 2010
  • 9. 9 How to meet the performance Objective ? Solution:  IMS may build multiple VSAM pools. Each pool may have multiple sub pools.  A VSAM pool is defined by a POOLID= statement in the DFSVSMxx member for online systems or the DFSVSAMP data set for batch and utility jobs.  Suggest Changes : POOLID=SP01 VSRBF=8192, 700, D DBD= XTDBD01 (1, SP01) POOLID=SP02 VSRBF=12288, 200, D VSRBF=10240, 500, I DBD= XTDBD02 (1, SP02) IOBF=(2048,4,Y,Y) OPTIONS, VSAMFIX= (BFR, IOB), BGWRT= (YES, 50) By dividing buffer pools in data and index component number of page hit ratio can be increased and reduce the I/O transfer rate/sec, which results in reduction of elapse time, CPU units, I/O transfer counts.  Buffer pool has been defined according to the VSAM CI or OSAM Block Size. AMS Share-Net | Oct 2010
  • 10. 10 How to meet the performance Objective ? Change Desc:Change Desc:  VSAM ESDS Buffer poolVSAM ESDS Buffer pool POOLID=SP01 /* POOLID: Define buffer pool name, SPO1: is buffer pool name. Its user define namePOOLID=SP01 /* POOLID: Define buffer pool name, SPO1: is buffer pool name. Its user define name VSRBF=8192, 700, D /*-VSRBF: Sub pools in the pool are defined by the VSRBF= statements which followVSRBF=8192, 700, D /*-VSRBF: Sub pools in the pool are defined by the VSRBF= statements which follow /*the POOLID= statement - 8192-: Buffer Size -Sub pool has a buffer size of 8192,/*the POOLID= statement - 8192-: Buffer Size -Sub pool has a buffer size of 8192, /*700 : Number of buffer pool/No. of block in SPO1 buffer pool./*700 : Number of buffer pool/No. of block in SPO1 buffer pool. D : Data component will use this buffer pool SP01.D : Data component will use this buffer pool SP01.  VSAM KSDS Buffer poolVSAM KSDS Buffer pool POOLID=SP02POOLID=SP02 VSRBF=12288,200,D / * D : Indicates that data component can use this subpoolVSRBF=12288,200,D / * D : Indicates that data component can use this subpool VSRBF=10240,500,I /* I : Indicated the only index component can use this subpool.VSRBF=10240,500,I /* I : Indicated the only index component can use this subpool. DBD=XTDBD02 (1,SP02) /* XTDBD02 : DBD name corresponding to dataset.DBD=XTDBD02 (1,SP02) /* XTDBD02 : DBD name corresponding to dataset. Job1 Shared Buffer Pools Job2 Online Scr Remember:Remember: Performance dependencePerformance dependence on whether shared bufferon whether shared buffer used by multiple programused by multiple program during run-timeduring run-time AMS Share-Net | Oct 2010
  • 11. 11 Test Results..  B&UT Testing : IBM GR Team – Test Results Description Base Converted Run1 CPU UNITS 3,026,234 2,843,993 CPU Time 00:01:14.88 00:01:02.37 SERV UNITS 3,092,247 2,906,908 SRB UNITS 51,595 48,524 I/O UNITS 14,418 14,391 Execution Time (step XNC100X) 00:10:55.09 00:05:30.62 DLI Statistics Report Calls Base Convert GU 47 41 GN 83,342 83,343 GNP 2,062,941 2,062,981 GHU 6 6 DB (TOTAL) 2,146,342 2,146,383 CHKP 42 42 LOG 294 294 Remember 1:Remember 1: Performance should bePerformance should be compared based on No. ofcompared based on No. of DLI calls issued & No. ofDLI calls issued & No. of record accessed inrecord accessed in program.program. AMS Share-Net | Oct 2010
  • 12. 12 Test Results..  SIT Testing : Client Technical Expert Group – Test Results Description Base Converted Run1 Converted Run2 Converted Run3 CPU UNITS 10,264,284 9,914,169 9,538,354 9,031,724 CPU Time 00:04:06.28 00:04:05.32 00:03:56.02 00:03:36.71 SERV UNITS 10,584,075 10,227,066 9,842,507 9,325,137 SRB UNITS 233,312 230,039 226,202 217,536 I/O UNITS 86,479 82,858 77,951 75,877 Execution Time (step XNC100X) 00:30:06.43 00:24:18.70 00:20:10.64 00:12:16.73 Remember 2:Remember 2: Whether the IMS1/IMS2Whether the IMS1/IMS2 region database has beenregion database has been re-org recentlyre-org recently AMS Share-Net | Oct 2010
  • 13. 13 Business Benefits…  Buffer pool concept will be applicable to batch & online application.  By proposing IMS buffer-pool concept we are able to reduce 10-20% of elapse time, CPU units, I/O transfer counts.  Implementing buffer-pool concept into there respective account will add values to clients and will be able to meet SLA, as it will reduce application elapse time.  Dividing buffer pools in data and index component number of page hit ratio can be increased and reduce the I/O transfer rate/sec  Supports additional new opts Included – BGWRT (Back Ground Write) % used for parallel write operation to database and buffer pool can be released during database processing. IMS BufferIMS IMS AMS Share-Net | Oct 2010
  • 14. 14 Performance Dependent Parameters…  Performance of XNC100JD dependent on the below parameters  Region Parameters: • Whether these database are used by other program during the batch job execution. • Shared buffer used by other program for loading the data. • Whether the IMS1/IMS2 region database has been reorganized recently.  Application Parameters: • No. of record retrieved from the database.. • No. of DLI call issued by the program. • BGWRT- The best percentage value to be specified will change depending on whether the program is re- accessing the buffers or not. • Background Write: In general, maximum benefit occurs in sequential update processing. AMS Share-Net | Oct 2010
  • 15. 15 Implementation Method…  Batch Jobs - Update the DFSVSAMP DD Control Cards  Online Program - DFSVSMxx member in IMS.PROCLIB for online systems Example 1: POOLID=VSM1 VSRBF=4096,3000,D VSRBF=12288,2000,I VSRBF=32768,12,D Since the second VSRBF statement builds an index subpool, the first and third subpools may only be used for data components. The second subpool may only be used for index components. Assumption:Assumption: Application is IMS BatchApplication is IMS Batch or Online Programor Online Program AMS Share-Net | Oct 2010
  • 16. 16 For more information  IMS Full Function Database Use of VSAM Buffer Pools -http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/FLASH10240  IMS Buffer Pool Analyzer - http://publib.boulder.ibm.com/infocenter/dzichelp/v2r2/index.jsp? topic=/com.ibm.imstools.bpl.doc.ug/bplucon_hilites.htm  IMS Performance Monitoring and Tuning Update -ibm.com/redbooks AMS Share-Net | Oct 2010
  • 17. 17 IMS Buffer Pool is … … Just Great ! THANK YOU …..ANY QUESTIONS AMS Share-Net | Oct 2010