ims shared queues - tuning and monitoring

52
ntroduction to Shared Queues and troduction to Shared Queues and oupling Facility Structures upling Facility Structures uning and Monitoring ning and Monitoring Jasdeep Singh IMS Performance [email protected] 1

Upload: ibm-ims

Post on 17-Feb-2017

289 views

Category:

Technology


15 download

TRANSCRIPT

Page 1: IMS Shared Queues - Tuning and Monitoring

Introduction to Shared Queues and Introduction to Shared Queues and Coupling Facility StructuresCoupling Facility StructuresTuning and MonitoringTuning and Monitoring

Jasdeep SinghIMS [email protected]

1

Page 2: IMS Shared Queues - Tuning and Monitoring

2

Agenda

Topics

Coupling Facility Structures

Lock, Cache, and List structures

IMS use of structures Explicit and implicit

Shared Queues Overview

• Components

What to tune XES or IMS

How to monitor

• RMF and IMS

Page 3: IMS Shared Queues - Tuning and Monitoring

3

Parallel Sysplex ComponentsSysplex Couple Data Set

Information about sysplex systems, XCF groups

Sysplex Timer

Synchronization of clocks on all systems

Coupling Facility (CF)

z Systems hardware with microcode

CF Channels

Special connections between the host and coupling facility

Sysplex Timer z/OS System

Coupling Facility

ESCON

z/OS System

DASD

CF Link

CF Link

112 2

345678910

11

Page 4: IMS Shared Queues - Tuning and Monitoring

4

CF Structures

Structures contain all of the user data in a CF

Structure types:

Cache

–Buffer coherency–Caching data

Lock

–Global locking services List

–Messages–State information–Data collection

Page 5: IMS Shared Queues - Tuning and Monitoring

5

Cache Services

Services provided

Storing and retrieving data (caching)

Registering interest in data item

Requesting invalidation of buffers containing a data item

Tracking "changed" and "unchanged" data items

Cache Structure Types

Directory Only

Store Through

Store In

Page 6: IMS Shared Queues - Tuning and Monitoring

6

Cache Structures

Directory Only

Contains no "data"

Used for local buffer coherency

Tracks which blocks from DASD are in which buffers in connectors (such as IMS)

–Buffers are associated with a bit in a "local vector" in HSA for the system

–Structure contains bit number for the vector

IMS uses these for VSAM buffer pools and optionally for OSAM

A

BD

E

F

G

a1 b4

a2a5

b7

b3

b2

Block from DASD

Buffer in connector 'a'

Buffer in connector ' b'

Page 7: IMS Shared Queues - Tuning and Monitoring

7

Cache Structures

Store-Through

Used for local buffer coherency

Contains "unchanged" data

–Same as on DASD–Modified buffers must still be written at syncpoint

OSAM may optionally use these

Store-In

Used for local buffer coherency

Contains "changed" data

–Different from DASD DEDB VSO use these

A

BD

E

F

G

a1 b4

a2a5

b7

b3

b2

Page 8: IMS Shared Queues - Tuning and Monitoring

8

Cache Services - Read and Register

When a connector reads a database block

Registers interest in block

–If block has no entry in cache structure, an entry is created

–If block already has an entry, it is updated

Once read, bit in HSA for this buffer is set to "valid"

If block in cache return data back to IMS

A

BD

E

F

G

x1 y4

x2x5

y7

y3

y2

Ax1

Connector X Connector Y

y4A

Page 9: IMS Shared Queues - Tuning and Monitoring

9

Cache Services - Read and Register

A

BD

E

F

G

x1 y4

x2x5

y7

y3

y2

Ax1

Connector X Connector Y

y4A

When a connector updates a block

Requests CF to invalidate buffers in other connectors containing the block

CF examines cache structure for other connectors which have the block in a buffer

CF sends signal to systems with those connectors

Receiving systems invalidate the buffers

–Bit in HSA is "flipped"For example

If IMSX were to update BlockA, IMSY's copy of BlockA would be invalid

Page 10: IMS Shared Queues - Tuning and Monitoring

10

Lock Services

Services provided

Determination of lock compatibility

Global contention resolution

Handling locks of failed systems

XCF group services used for communication of contention information

Page 11: IMS Shared Queues - Tuning and Monitoring

11

Lock StructureLock Structure has two parts:

Lock Table (Hash Table) - Used to grant locks

• Used to track which lock managers have potential interest in a lock

• Locked resources are hashed to entry

• Each entry indicates which lock managers have requested lock on a resource that hashes to entry

• Record List – Used for recovery• Lock manager may store locks in

this list for recovery purposes• If lock manager fails, partner lock

managers have access to these locks

Lock structure

Lock table

Record list

Lock Mgr. 1 Lock Mgr. 2 Lock Mgr. 3

Page 12: IMS Shared Queues - Tuning and Monitoring

12

Lock Services

Lock manager (e.g. IRLM) uses lock services

Lock services keeps locks in data spaces

Lock services accesses lock structure

Lock services uses XCF to communicate with other lock services

Communication required when lock table shows potential conflict

IRLM

Lock Services Locks

IRLM

Lock Services

Locks

XCFLock Structure

Page 13: IMS Shared Queues - Tuning and Monitoring

13

List Services

Services provided

Keeping of state information and data

Passing messages

Collecting data

Page 14: IMS Shared Queues - Tuning and Monitoring

14

List Structures

List Structure

Lock Table (optional)

–Used for serialization List Headers

–Anchors each list in structure List Entry Controls

–Control info. for entries in lists–Optionally point to data elements–Entries with same key form sublist

Data elements

–Hold user data Event Monitor Controls (optional)

–Contain information about sublists Event Queue Controls (optional)

–One for each connector

LockTable

ListHeaders

List EntryControls

Data Elements

Event QueueControls

Event Monitor Controls

Page 15: IMS Shared Queues - Tuning and Monitoring

15

List StructuresUsed by IMS (CQS) for

•Primary and Overflow message queue structures

•Primary and Overflow EMH (FP) structures

•RM Resource Structure

Used by z/OS logger for logstreams

Page 16: IMS Shared Queues - Tuning and Monitoring

16

Coupling Facility Resource ManagementManages CF resources

CFRM policy defines CFs in the Parallel Sysplex

CFRM policy defines which structures may be built

–Names of structures–Sizes of structures - can use CFSIZER on IBM web site

http://www.ibm.com/servers/eserver/zseries/cfsizer–CFs which are candidates to hold a structure

CFRM couple data set contains

–CFRM Policies, only one active policy in sysplex–Status Data

Current structuresConnectors to current structures

–Think of like RECONs, not highly used but critical to performance

Page 17: IMS Shared Queues - Tuning and Monitoring

17

What are Shared Queues?

A set of input and output message queues which can be shared by multiple IMSs in a Parallel Sysplex.

-Full Function Queue-Fast Path Queue

ListHeaders Messages

QUEUES

IMS1

List Structure in Coupling Facility

IMS2

IMS3

Page 18: IMS Shared Queues - Tuning and Monitoring

18

Without Shared MSG/BALG

IMSs could share the data, but each IMS had exclusive use of its own message queues

Message Queue Dataset and Fast Path Balancing Group which is in memory

If one IMS fails, any messages in its queue are not processes until restarted

Application load balancing among IMS systems was a user responsibility

Network balancing, Workload Router, MSC, ISC, APPC, ....

IMSnIMS2IMS1

SharedData

IMS1MessageQueues

IMSnMessageQueues

IMS2MessageQueues

BALG BALG BALG

Page 19: IMS Shared Queues - Tuning and Monitoring

19

With Shared MSG/BALG

Individual IMS message queues have been replaced by Shared Queue Structures in the Coupling Facility

Messages in the Shared Queue Structures are available to all IMSs for processing

Coupling Facility

IMS1 IMSnIMS2

TRAN

LTERM LTERM1

TRANX TRANX TRANY

LTERM2

SharedQueueStructure

SharedData

Page 20: IMS Shared Queues - Tuning and Monitoring

20

Queues are maintained in List Structures in the Coupling Facility

Structure includes a list header pointing to input messages and another header pointing to output messages

Defined in CFRM Policy

One primary structure for FF messages

–Optional overflow structure One primary structure for EMH messages

–Optional overflow structure

Shared Queues

Coupling Facility

ListStructu

re

Transaction Queue

LTERM Queue

ListHeaders

List Entries(Messages)

Other Queues

TRAN

LTERM

Q-TYPE

MSG1 MSG2 MSG3

MSG1

MSG1 MSG2

Page 21: IMS Shared Queues - Tuning and Monitoring

21

IMS (CQS component) : Handles interaction between IMS and CF

Connects to MSGQ and to EMHQ list structures

–Connection is through a Common Queue Server (CQS)

Registers interest in specific queues–Indicates that IMS is capable of processing a message on that queue

Implementation of Shared Queues

IMSA CQSB IMSBCQSA CF

MSGQ StructuresEMHQ Structures

IMSA CQSB IMSBCQSACF

TRANXTRANYLTERM1

TRANXTRANZLTERM2

Page 22: IMS Shared Queues - Tuning and Monitoring

22

Implementation ...

Registering interest

IMS registers interest in queue names it is able to process

–Input transactions (can schedule)–Output messages (can deliver)

Message handling

When IMS receives an input message it places it on a shared input

(FF transaction or FP program) ready queue

–Any IMS with registered interest in the transaction may retrieve it from the shared queue and process it

–May be multiple IMSs with registered interest When IMS has an output message, it places it on a shared output (e.g. LTERM) ready

queue–Any IMS with registered interest in the LTERM may retrieve it from the queue and send it

–Should be only one IMS with registered interest

Page 23: IMS Shared Queues - Tuning and Monitoring

23

Shared Queues Components

Coupling facility structures

Message Queue Structures (Primary and Overflow)

EMH Queue Structures (Primary and Overflow)

The EMHQ structure has been optional since IMS V9 MVS Logger Structures (MSGQ and EMHQ log streams)

Common Queue Server (CQS) address space

Possible to have multiple IMS per CQS Not really recommended for production

environments Interface between IMS and shared queue structures

CQS Checkpoint Data Sets

One per queue structure for each CQS Used for CQS restart

Coupling Facility

MSGQ

EMHQ

MSG LOGEMH LOG

CQS IMSCTL

DLISAS

DBRC

CHKPTCHKPT

Page 24: IMS Shared Queues - Tuning and Monitoring

24

Shared Queues Components ...

Structure Recovery Data Sets

One pair per queue structure Shared among all CQSs that share a queue structure Used, with z/OS logstream, to recover

structuresz/OS Log Streams

One per queue structure CQS logs MSGQ and EMHQ activity

into logstreams–All CQSs use the same logstream(s)

Logstreams implemented as –List Structures–Structure offload (spill) data sets–Staging data sets (optional)

Used for structure recovery and CQS restart

CQS1 CQS2

CQS3

MSG LOG

EMH LOG

MSGQ

EMHQ

SRDS2

SRDS1

SRDS2

SRDS1

MSG

EMH

CHECKPOINT DATA SETS

LOGGERDATA SETS

Page 25: IMS Shared Queues - Tuning and Monitoring

25

Shared Queues Components ...

Common CQS Data Sets

MSGQ Logger Structure

Coupling Facility

EMHQCHKPT MSGQ

CHKPTMSGQCHKPT

EMHQCHKPT

MSGQSRDS2MSGQ

SRDS1

EMHQSRDS2EMHQ

SRDS1

MVSLOGLOGGER

SPILL

MVSLOGLOGGERSTAGING

MVSLOGLOGGERSTAGING

EMHQ Logger Structure

MSGQ OFLW StructureMSGQ Primary Structure

EMHQ OFLW StructureEMHQ Primary Structure

IMS2CQS1

LOGGER

LOGGER

CQS2

IMS1

OLDS OLDS

Page 26: IMS Shared Queues - Tuning and Monitoring

26

Benefits of Shared Queues

Automatic work load balancing

A message placed on the Shared Queues can be processed by any IMS with interest in the message

Incremental growth

New IMS subsystems can be added as workload increases

New IMSs can be for processing only (no network) during periods of heavy activity

Improved availability

If an IMS fails, the workload may be assumed by the surviving IMSs

Shared queues are not lost if one or more IMSs are cold started

Page 27: IMS Shared Queues - Tuning and Monitoring

27

Shared Queues PROCLIB Members

DFSPBxxx: IMS execution parameters

SHAREDQ=xxx

Buffer allocation parameters

DFSSQxxx: contains information for connecting to CQS

CQSIPxxx: CQS initialization parameters

CQSSLxxx: CQS local structure definitions

CQSSGxxx: CQS global structure definitions

Page 28: IMS Shared Queues - Tuning and Monitoring

28

IMS/CQS Library Relationships

CQSA

DFSSQ00A

CQSIP00A

CQSSL00A

CQSSG001

CF

CDS

RESLIB

DFSSQ00BIMSBIMSA

PROCLIB PROCLIB

PROCLIBCQSIP00B

CQSSL00BCQSB

Page 29: IMS Shared Queues - Tuning and Monitoring

29

Data object storage

RDY

STG

ListHeaders Controls/Adjunct Areas Data Elements

LockTable

List Entries

LE 1

LE 1

LE 3

LE 2

LE 2

LE 2

LE 3

LE 3

MSG1 = 1 DATA OBJECT (LE1) = 1 LIST ENTRY CONTROL + 1 DATA ELEMENTMSG2 = 2 DATA OBJECTS (LE2+LE3) = 2 LIST ENTRY CONTROLS + 4 DATA ELEMENTSDefine the ratio of List Entries to Data Elements in IMS

Message Queue Structure

MSG1

MSG2

MSG2

Page 30: IMS Shared Queues - Tuning and Monitoring

Shared Queues Considerations

Overflow ProcessingOperations procedures

30

Page 31: IMS Shared Queues - Tuning and Monitoring

31

PrimaryStructure

OverflowStructure

When first invoked, an alter of the structure is initiated.When Primary Structure reaches a user-defined threshold (percent full), 20% of Primary Structure may be moved to Overflow Structure.

Structure Overflow

A queue can not be in both primary and overflow structure, it is recommended to have overflow structure slightly larger than the primary structure.

Page 32: IMS Shared Queues - Tuning and Monitoring

32

Overflow Processing

Each message PUT on the shared queue uses ...

One list entry control (long messages may require more)

One or more 512-byte data elements

–Contains IMS Message Prefixes plus User DataIf we run out of either list entry controls or data elements

Structure is FULL

PUT requests are rejected until condition is relieved

IMS only monitors data elements

This condition may occur because ...

Structures are grossly undersized

Some queue destinations are not receiving

–Transaction/LTERM/MSC Link is stopped–Printer is out of paper/powered off/slow–Destination doesn't exist, but Output Creation Exit sent it there anyway

Application program errors or loops

Page 33: IMS Shared Queues - Tuning and Monitoring

33

List Structures (A Reminder)

List structure resources

RDY

STG

ListHeaders List Entry Controls Data

Elements

LockTable

List Entries

LE 1

LE 1

LE 3

LE 2

LE 2

LE 2

LE 3

LE 3

MSG1 = LE1MSG2 = LE2 + LE3

IMS keeps track of data elements to trigger overflow processing.Specify good ratio of LE to data elements

Page 34: IMS Shared Queues - Tuning and Monitoring

34

Overflow Processing ...

Overflow processing is intended to isolate the offenders

Invoked when percentage of data elements in use reaches a user-defined threshold percentage–Defined in PROCLIB member CQSSGxxx–Default: OVFLWMAX=70%

For structures defined with INITSIZE parameter in the CFRM Policy–If structure size is less than maximum SIZE, structure size is dynamically increased to bring utilization down 20%

Queues (QNAMEs) using the most data elements are identified as candidates for overflow processing–Structure activity is quiesced while candidates are selected–Candidate list represents QNAMEs using 20% of total data elements in primary structure

–Maximum of 512 candidate QNAMEs User exit can override CQS decision to put QNAME in overflow status

–Exit name defined in library member at start-up time

Page 35: IMS Shared Queues - Tuning and Monitoring

35

Overflow Processing ...

/CQQUERY STATISTICS STRUCTURE structurename|ALL

Obtains statistical information regarding current utilization of shared queue structures

If ELMINUSE reaches 10500 overflow processing will be invoked

–Assuming default overflow threshold of 70% If LEINUSE reaches 5000 structure will be FULL

–Overflow processing does not detect shortage of List Entries

STRUCTURE NAME LEALLOC LEINUSE ELMALLOC ELMINUSE LE/EL

MSGQPRIMARY 5000 4000 15000 5000 1/3

LEALLOC Number of list entries allocated. Maximum number of messages possible.

LEINUSE List entries in use. Number of messages currently on queue.ELMALLOC Number of 512-byte list elements allocated.ELMINUSE Number of list elements in use. Drives overflow processing.LE/EL Ratio of list entries to list elements.

Page 36: IMS Shared Queues - Tuning and Monitoring

36

What can you do about it?

Several commands available to monitor and correct

–/DIS STRUCTURE ALLIndicates whether primary structure is in overflow mode

–/DIS OVERFLOWQ STRUCTURE ALLIdentifies QNAMEs in overflow

–/DIS QCNT LTERM xxxxxxxx MSGAGE 0Indicates how many messages are on queue xxxxxxxx

–/DEQ TRANSACTION|LTERM|MSNAME xxxxxxxx PURGE1|FIRSTPurges 1st message for QNAME xxxxxxxx

– /DEQ TRANSACTION|LTERM|MSNAME xxxxxxxx PURGEPurges all messages for QNAME xxxxxxxx

If Queues Are Getting Too Full

Page 37: IMS Shared Queues - Tuning and Monitoring

37

If Queues Are Getting Too Full ...

Can make structure larger

Change CFRM Policy

–Specify a larger SIZE

Enter MVS command to activate policy

–SETXCF START,POLICY,TYPE=CFRM,POLNAME=POLICY2–Structure change will be pending until structure is rebuilt

Enter MVS command to rebuild structure

–SETXCF START,REBUILD,STRNAME=MSGQSTR–Structure will be rebuilt with new size specification–Structure is quiesced during rebuild

Page 38: IMS Shared Queues - Tuning and Monitoring

38

Structure Full Condition

Either primary or overflow structure may become full

Out of data elements Out of list entries

If either structure is full

PUTs to that structure are rejected READs are allowed

Structure full is usually a temporary condition

DELETEs free up space IMS does not stop trying to PUT messages on a full structure

When structures are full, IMS does not abend with a U758. It continues to attempt to process messages.

Page 39: IMS Shared Queues - Tuning and Monitoring

39

Structure RebuildStructures may be rebuilt while in use

Rebuild may be result of:

–operator command SETXCF START,REBUILD,STRNM=strname

–failure of structure, CF, or connection Rebuild requires code in connectors

–Some connectors do not support rebuild–Connectors which support rebuild may work differently

Some restore dataSome build empty structure

MVS merely supervises rebuild

If not in use (no connectors)

System managed rebuild may be initiated by command and does not require connector participation

CF CF

MVS MVS

Page 40: IMS Shared Queues - Tuning and Monitoring

40

Structures may be altered in place

Alter changes size or internal

characteristics of structure

Alter may be result of:

–operator commandSETXCF START,ALTER,STRNM=strname,SIZE=size–request from connector–automatically by system (autoalter)

Alter capability is optional for a structure

–Specified when built Connectors do not participate in alter

process

Max size limited by CFRM policy SIZE parm

Structure Alter

CF CF

MVS MVS

Page 41: IMS Shared Queues - Tuning and Monitoring

41

Performance

Performance Components

Processor power under z/OS system

Subchannel availability within each z/OS system

IOP

Physical path availability

CF link speed

CF processing power

Structure attributes

–Size–Usage

Page 42: IMS Shared Queues - Tuning and Monitoring

42

Performance

Performance Considerations

Unavailability of resources leads to elongated response times

Response time composed of:

–Delay Time + Service Time Delay time is spent obtaining a subchannel

–May be reflected in CPU busy (depends on request type) Service Time reflects time from MVS CF command operation started to completion

–Multiple components (i.e. CF Link Speed, CF speed, CF busy)–May be reflected in CPU busy (depends on request type)

Page 43: IMS Shared Queues - Tuning and Monitoring

43

Request modes:

Synchronous

–Requester waits for operation to complete–Delay Time and Service Time are reflected in CPU busy time

Asynchronous

–Requester does not wait for operation to complete–Processor is freed to do other work–Delay Time and Service Time are not reflected in CPU busy time

Request mode may be determined by requester

–Some requests allow only one of the modes–Some requests may be either synchronous or asynchronous–Some synchronous requests are converted to asynchronous by XES

XES has internal algorithm to determine

Performance

Page 44: IMS Shared Queues - Tuning and Monitoring

44

RMF reports on CF

Coupling Facility Usage Summary

–Storage allocation and usage–Structure activity–CF processor utilization

Coupling Facility Structure Activity

–System level detail by structure–Request counts and rates by structure–Service and queue times by structure

Coupling Facility Subchannel Activity

–Activity summary by system –Path/Channel busy counts–Requests counts, rates, service, and queue times by system

RMF

Page 45: IMS Shared Queues - Tuning and Monitoring

45

Number of Lock Requests affects transaction performance

Even without data sharingBut structure access adds to overheadReduce locking requests

Use Procopt=GO where possible

Minimize Contention

False contention

•Increase structure size

•Usually best to increase by power of 2

•There is a point of diminishing return howeverReal contention

•Application and/or DB design

IMS Lock Structure

Page 46: IMS Shared Queues - Tuning and Monitoring

46

RMF Lock Structure report

STRUCTURE NAME = IRLMLOCKTBL1 TYPE = LOCK STATUS = ACTIVE # REQ -------------- REQUESTS ------------- -------------- DELAYED REQUESTS ------------- SYSTEM TOTAL # % OF -SERV TIME(MIC)- REASON # % OF ---- AVG TIME(MIC) ----- EXTERNAL REQUEST NAME AVG/SEC REQ ALL AVG STD_DEV REQ REQ /DEL STD_DEV /ALL CONTENTIONS STLAB77 1859K SYNC 1859K 100 1.8 0.9 NO SCH 0 0.0 0.0 0.0 0.0 REQ TOTAL 2266K 15494 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 REQ DEFERRED 176 CHNGD 0 0.0 INCLUDED IN ASYNC PR CMP 0 0.0 0.0 0.0 0.0 -CONT 176 SUPPR 0 0.0 -FALSE CONT 86 ----------------------------------------------------------------------------------------------------------------------------- TOTAL 1859K SYNC 1859K 100 1.8 0.9 NO SCH 0 0.0 0.0 0.0 0.0 REQ TOTAL 2266K 15494 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 REQ DEFERRED 176 CHNGD 0 0.0 PR CMP 0 0.0 0.0 0.0 0.0 -CONT 176 SUPPR 0 0.0 -FALSE CONT 86

Page 47: IMS Shared Queues - Tuning and Monitoring

47

IMS Cache Structures• OSAM & VSAM

– IMS will still write modified data to DASD at syncpt• OSAM

– Use data caching only for good reason• Overhead may not be worth reduced I/O• DIR to ELE ratio of 1:0 for no caching

– Avoid directory reclaims (aka:castouts)• Not enough entries to accommodate all buffers

• VSAM– Avoid directory reclaims

• FP VSO– If Preload option make size appropriate

Page 48: IMS Shared Queues - Tuning and Monitoring

48

RMF Cache Structure report

STRUCTURE NAME = OSAMSESXI1 TYPE = CACHE STATUS = ACTIVE # REQ -------------- REQUESTS ------------- -------------- DELAYED REQUESTS ------------- SYSTEM TOTAL # % OF -SERV TIME(MIC)- REASON # % OF ---- AVG TIME(MIC) ----- NAME AVG/SEC REQ ALL AVG STD_DEV REQ REQ /DEL STD_DEV /ALL STLAB77 698K SYNC 698K 100 1.8 0.5 NO SCH 0 0.0 0.0 0.0 0.0 5820 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 CHNGD 0 0.0 INCLUDED IN ASYNC PR CMP 0 0.0 0.0 0.0 0.0 SUPPR 0 0.0 DUMP 0 0.0 0.0 0.0 0.0 -----------------------------------------------------------------------------------------------------------------------------TOTAL 698K SYNC 698K 100 1.8 0.5 NO SCH 0 0.0 0.0 0.0 0.0 -- DATA ACCESS --- 5820 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 READS 0 CHNGD 0 0.0 PR CMP 0 0.0 0.0 0.0 0.0 WRITES 203167 SUPPR 0 0.0 DUMP 0 0.0 0.0 0.0 0.0 CASTOUTS 0 XI'S 150378

STRUCTURE NAME = VSAMSESXI1 TYPE = CACHE STATUS = ACTIVE # REQ -------------- REQUESTS ------------- -------------- DELAYED REQUESTS ------------- SYSTEM TOTAL # % OF -SERV TIME(MIC)- REASON # % OF ---- AVG TIME(MIC) ----- NAME AVG/SEC REQ ALL AVG STD_DEV REQ REQ /DEL STD_DEV /ALL STLAB77 1183K SYNC 1183K 100 1.9 2.1 NO SCH 0 0.0 0.0 0.0 0.0 9862 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 CHNGD 0 0.0 INCLUDED IN ASYNC PR CMP 0 0.0 0.0 0.0 0.0 SUPPR 0 0.0 DUMP 0 0.0 0.0 0.0 0.0 -----------------------------------------------------------------------------------------------------------------------------TOTAL 1183K SYNC 1183K 100 1.9 2.1 NO SCH 0 0.0 0.0 0.0 0.0 -- DATA ACCESS --- 9862 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 READS 0 CHNGD 0 0.0 PR CMP 0 0.0 0.0 0.0 0.0 WRITES 0 SUPPR 0 0.0 DUMP 0 0.0 0.0 0.0 0.0 CASTOUTS 0 XI'S 2058635

Page 49: IMS Shared Queues - Tuning and Monitoring

49

IMS List Structures• Message Structures (FF & FP)

– Minimize structure access• Select QBUFSZ, LGMSGSZ, SHMSGSZ

carefully– The more messages that fit in a single QBUF the better

– Avoid full structure and rebuilds• Appropriate size structure

– Allow for processing delays• Overflow structures

– Larger than primary• Directory(entry) to element ratio (OBJAVGSZ)

Page 50: IMS Shared Queues - Tuning and Monitoring

50

RMF List Structure report

STRUCTURE NAME = IMSMSGQ01 TYPE = LIST STATUS = ACTIVE # REQ -------------- REQUESTS ------------- -------------- DELAYED REQUESTS ------------- SYSTEM TOTAL # % OF -SERV TIME(MIC)- REASON # % OF ---- AVG TIME(MIC) ----- EXTERNAL REQUEST NAME AVG/SEC REQ ALL AVG STD_DEV REQ REQ /DEL STD_DEV /ALL CONTENTIONS STLAB77 927K SYNC 927K 100 2.7 1.0 NO SCH 0 0.0 0.0 0.0 0.0 REQ TOTAL 927K 7725 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 REQ DEFERRED 0 CHNGD 0 0.0 INCLUDED IN ASYNC PR CMP 0 0.0 0.0 0.0 0.0 SUPPR 0 0.0 DUMP 0 0.0 0.0 0.0 0.0 ----------------------------------------------------------------------------------------------------------------------------- TOTAL 927K SYNC 927K 100 2.7 1.0 NO SCH 0 0.0 0.0 0.0 0.0 REQ TOTAL 927K 7725 ASYNC 0 0.0 0.0 0.0 PR WT 0 0.0 0.0 0.0 0.0 REQ DEFERRED 0 CHNGD 0 0.0 PR CMP 0 0.0 0.0 0.0 0.0 SUPPR 0 0.0 DUMP 0 0.0 0.0 0.0 0.0

Page 51: IMS Shared Queues - Tuning and Monitoring

51

IMSPA MSGQ report

Message Queue Utilization-IM02 ______________________________ Msg Length Msg Avg Input Transaction -Message Switch-- -Program Switch-- -Output Message-- ------Totals----- Acc Interval Length Count ShMsg LgMsg Count ShMsg LgMsg Count ShMsg LgMsg Count ShMsg LgMsg Count ShMsg LgMsg Pct Pct ___________ _______ _________________ _________________ _________________ _________________ _________________ ____ ____ 00400-00499 491 - - - - - - - - - 34140 34140 - 34140 34140 - 9 9 00500-00599 567 6678 6678 - - - - - - - 10 10 - 6688 6688 - 2 11 00700-00799 734 - - - - - - - - - 15077 15077 - 15077 15077 - 4 16 00800-00899 815 9895 9895 - - - - - - - - - - 9895 9895 - 3 18 00900-00999 930 - - - - - - - - - 6653 6653 - 6653 6653 - 2 20 01200-01299 1216 99468 29531 69937 - - - - - - 8157 8157 - 107K 37688 69937 30 50 01300-01399 1379 8171 16342 - - - - - - - 28636 - 28636 36807 16342 28636 10 60 ____________________________________________________________________________________________________________________________________01400-01499 1453 8157 16314 - - - - - - - - - - 8157 16314 - 2 62 01600-01699 1656 - - - - - - - - - 28638 39198 9039 28638 39198 9039 8 70 02100-02199 2155 13265 - 13265 - - - - - - 6676 13352 - 19941 13352 13265 6 76 02200-02299 2256 8381 - 8381 - - - - - - 8172 - 8172 16553 - 16553 5 81 02300-02399 2324 - - - - - - - - - 22183 - 22183 22183 - 22183 6 87 02700-02799 2728 - - - - - - - - - 16187 - 16187 16187 - 16187 4 91 02800-02899 2894 - - - - - - - - - 6628 - 6628 6628 - 6628 2 93 03000-03099 3085 9108 - 9108 - - - - - - - - - 9108 - 9108 3 96 03500-03599 3528 - - - - - - - - - 16158 - 16158 16158 - 16158 4 100 ____________________________________________________________________________________________________________________________________Total 1573 163K 78760 100K - - - - - - 197K 116K 107K 360K 195K 207K -/- -/- Est. Short Message Queue Dataset Record Size is 1305 Est. Long Message Queue Dataset Record Size is 3915

Page 52: IMS Shared Queues - Tuning and Monitoring

52

Thank You