ims12 ims performance tools
TRANSCRIPT
© 2012 IBM Corporation 1 © 2015 IBM Corporation
IMS Technical Conference Australia 2015
IMS12: Are you getting the most out of your IMS performance tools?
John Hancy, Fundi Software
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
2
Agenda
Useful new information available in the IMS log
Techniques for gaining insight from this new
information
Recent reporting improvements that you might find
useful
The IMS transaction index – what is it and what
can I do with it?
Interactive problem determination using the IMS
log and other data sources – a live demo
Not just for IMS – perhaps your colleagues might
find these tools useful too?
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
3
What customers are asking for
1. Total Cost of Ownership – CPU measurement,
apportionment (per subsystem), reduction
Mobile Workload Pricing (MWP) - see IMS newsletter
page 5
2. Providing application developers with relevant information
to tune their applications, including:
3. Mobile and distributed – what’s happening on z/OS?
4. Batch window optimization
5. Bridging CICS, DB2, MQ and IMS performance and
problem determination – connecting SME’s
6. Impact of software upgrades, new functions (sync callout
etc.) and application changes
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
4
The IMS log: what’s new or been hidden from view? Information my manager might be interested in
– Transaction workload rates coupled with response time and service levels
Database contention
– Lock wait time per transaction
– Maximum locks held per transaction
Database activity
– DB I/O count and elapsed time
– VSAM and OSAM read and write counts
Syncpoint
– Phase 1 and 2 breakdown
– Fast Path OTHREAD – post-syncpoint AREA update activity
External subsystem usage (DB2, MQ etc.)
– Subsystems used and the number of calls made
Shared queues
– Compare local-first versus global input queue times
Program switch
– Track the flow and delay in a program switch sequence
Synchronous call-out
– Elapsed time per call
BMP
– Checkpoint intervals, databases updated and used
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
5
LOGINFO – what is in your IMS log? Code Count MCNT Recs/Sec Avg len Max Len Byte/Sec MB %
01 IN 20,778 25 1,236 9,608 31,144 25.6 2.1 IMS Message
CQS PUT 20,059 24 1,229 9,608 29,899 24.6 2.0 Put onto shared queue
CQS RD 636 0 1,072 1,155 826 0.6 0.1 Read off shared queue
01 OUT 40 0 751 1,040 36 0.0 0.0 IMS Message
MSG SWI 40 0 751 1,040 36 0.0 0.0 Message switch
03 IN 2,398 2 1,214 9,350 3,530 2.9 0.2 IMS Message
CQS PUT 2,037 2 1,154 5,482 2,850 2.3 0.2 Put onto shared queue
CQS RD 187 0 537 1,155 121 0.1 0.0 Read off shared queue
03 OUT 20,478 24 1,664 9,608 41,325 34.0 2.8 IMS Message
CQS PUT 19,623 23 1,715 9,608 40,798 33.6 2.8 Put onto shared queue
CQS RD 29 0 481 1,008 16 0.0 0.0 Read off shared queue
MSG SWI 826 1 510 9,589 510 0.4 0.0 Message switch
07 1,417 25,953 1 456 456 783 0.6 0.1 Program schedule end
MPP 1,309 1,645 1 456 456 723 0.5 0.0 MPP
QUICK 67 24,300 0 456 456 37 0.0 0.0 MPP quick reschedule
FALSE 23 0 0 456 456 12 0.0 0.0 MPP false schedule
BMP 16 0 0 456 456 8 0.0 0.0 BMP
ABEND 2 8 0 456 456 1 0.0 0.0 Abended transaction
5921 1 0 160 160 0 0.0 0.0 FP DEDB ADS open
5937 16,838 20 202 202 4,122 3.4 0.3 FP syncpoint
5938 2 0 202 202 0 0.0 0.0 FP syncpoint failure
5950 388,059 470 2,276 8,313 1,070,760 883.3 72.5 FP database update
5954 1 0 4,230 4,230 5 0.0 0.0 FP DB update - DEDB open
5955 1,087 1 49 49 64 0.0 0.0 FP SDEP CI buffer obtained
5957 387 0 157 338 73 0.0 0.0 FP DB update - DMAC
5958 1,087 1 49 49 64 0.0 0.0 FP current SDEP CI write
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
6
Ranking transaction performance
Avg Avg Max >2.0 Tot Avg Avg Tran InputQ Process Process Process Process CPU DB Call Trancode Count Time Time Time Time Time Time Count STOCK 275 0.595132 3.569668 69.50787 12.00% 89.24169 0.095634 1 BALANCE 152 0.138350 2.858584 5.738182 50.00% 11.43433 0.044236 3 WITHDRAW 339 1.264292 2.517168 72.50948 10.26% 98.16957 0.049611 190 ORDERS 101 0.118796 2.515761 4.832235 27.20% 10.06304 0.041714 3 BANKING 147 0.024561 1.913816 2.492869 66.30% 3.827631 0.056932 2
Avg Avg Max >2.0 Tot Avg Avg Tran InputQ Process Process Process Process CPU DB Call Trancode Count Time Time Time Time Time Time Count WITHDRAW 339 1.264292 2.517168 72.50948 10.26% 98.16957 0.049611 190 STOCK 275 0.595132 3.569668 69.50787 12.00% 89.24169 0.095634 1 BALANCE 152 0.138350 2.858584 5.738182 50.00% 11.43433 0.044236 3 BANKING 147 0.024561 1.913816 2.492869 66.30% 3.827631 0.056932 2 ORDERS 101 0.118796 2.515761 4.832235 22.70% 10.06304 0.041714 3
Questions your management might want answers to…
1. Which high volume transactions are taking longest to process?
2. What percentage of high volume transactions failed to meet their two
second service level?
Use the information to target specific transactions for attention
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
7
Avg Avg Avg Min >0.001 Max Tran CPU Process Schedule Schedule Schedule Schedule Schedule Tran Count Time Time Time Time Required Time Time CHECKING 674 0.002497 0.004434 0.000163 0.000065 99.11% 0.15% 0.165942 CREDIT 335 0.003022 0.064634 0.000159 0.000075 78.21% 0.38% 0.007718 FINANCE 1,330 0.002205 0.010187 0.000153 0.000069 78.95% 0.19% 0.004631 MOBILE 2,101 0.000747 0.012407 0.000221 0.000106 5.33% 1.79% 0.004609 ORDER 1,645 0.001053 0.007522 0.000162 0.000096 60.55% 0.10% 0.003647 STOCK 1,681 0.001364 0.002444 0.000073 0.000034 99.82% 0.24% 0.003446 WITHDRAW 1,524 0.001207 0.007537 0.000068 0.000030 100.00% 0.07% 0.007919
Scheduling and Pseudo WFI
Transaction perspective
Dependent region perspective
Subsystem perspective
Avg Avg Avg Min >0.001 Max Tran CPU Process Schedule Schedule Schedule Schedule Schedule Region Count Time Time Time Time Required Time Time IBANKING 2,797 0.002384 0.015831 0.000145 0.000030 99.93% 0.11% 0.165942 IMOBILE 5,082 0.001804 0.025657 0.000143 0.000040 46.79% 0.21% 0.004631 IORDERS 2,679 0.002510 0.019839 0.000137 0.000031 99.93% 0.19% 0.048253 IWEB 1,111 0.001423 0.021483 0.000172 0.000039 8.46% 1.06% 0.004609
Avg Avg Avg Min >0.001 Max Tran CPU Process Schedule Schedule Schedule Schedule Schedule SSID Count Time Time Time Time Required Time Time IMSP 182,440 0.004517 0.034590 0.000119 0.000027 51.54% 0.21% 0.165942
Percentage of
transactions that
required a full
program schedule
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
8
Syncpoint – affect on response time
Process CPU DB Syncpt Phase 1 Phase 2 Phase 2 OTHREAD Trancode Count Time Time Calls Time Time Time FP Time Time BANK1 18 0.041045 0.006635 8 0.014769 0.000030 0.014739 0.014739 0.000980 BANK2 26 0.051985 0.005922 8 0.000071 0.000032 0.000039 0.000039 0.001107 BANK3 280 0.048873 0.004889 23 0.000038 0.000018 0.000019 0.000019 0.001203 ORDERS 299 0.044485 0.004687 23 0.000034 0.000017 0.000017 0.000017 0.001200 MOBILE1 316 0.040288 0.004833 23 0.000789 0.000019 0.000771 0.000028 0.000789 MOBILE2 307 0.057567 0.004862 23 0.000037 0.000018 0.000019 0.000019 0.001195 MOBILE3 282 0.033802 0.004694 23 0.000036 0.000019 0.000016 0.000016 0.001191 WEBORDER 19 0.039227 0.007359 9 0.000076 0.000034 0.000043 0.000043 0.001057 STOCK1 21 0.036017 0.006329 9 0.000070 0.000031 0.000039 0.000039 0.001034 STOCK2 15 0.299056 0.020897 21 0.000053 0.000036 0.000016 0.000016 0.000983 STOCK3 16 0.372236 0.020957 21 0.000056 0.000039 0.000017 0.000017 0.001064
Are you using external subsystems and DASD mirroring? These might
affect the time it takes for your transactions to commit: • Total syncpoint time
• Phase 1 and Phase 2
• Phase 2 attributable to Fast Path database only
• OTHREAD time to write to ADS (asynchronous – not part of SYNCPT)
Avg Avg Avg Avg Avg Avg Tran CPU Process Syncpt Sync PH1 Sync PH2 Sync PH2 SSID Count Time Time Time Time Time FP Time DB2P 47,764 0.005166 0.034179 0.005902 0.006919 0.000319 0.000049 None 91,899 0.001711 0.021559 0.001961 0.002202 0.000221 0.000204
What is the impact
of DB2 syncpoint?
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
9
Avg Avg Max Max Tran Process DB Call DB Lock HWM Trancode Count Time Count Time Locks NIGHTRUN 2 20.21559 208,528 3.014112 38,275 CHECING 5,585 0.034276 53 0.697122 62 CREDIT 1,160 0.145987 115 0.030345 63 FINANCE 129 0.012458 7 0.015962 25 MOBILE 7 0.324563 989 0.014670 294 ORDER 397 0.186213 59 0.013845 62 STOCK 48 0.079283 213 0.009023 33
Locking and database I/O
Locks taken
Database I/O performance and activity
Avg Avg Tot Tot Tot Tot DB IO DB IO OSAMRead OSAMWrit VSAMRead VSAMWrit Count Time Count Count Count Count 17,282 7.972630 0 0 30,722 3,841 21 0.017631 1,980 2 108,248 8,921 242 0.119385 171 0 18,243 1,145 22 0.010127 0 0 2,654 0 758 0.189440 0 0 5,307 0 229 0.120370 24 0 2,950 226 76 0.016838 0 0 3,662 0
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
10
Start 01Jun2014 12:04:23:01 IMS Performance Analyzer Database Update Activity-IMSP _____________________________ 5050 Database Program Proc Total Updates ISRT DLET REPL ________ ________ ____ _________ ________ ________ ________ ________ ___ ORDERS ONLINE APPL 152 DLI 152 38 76 38 I/O 114 38 0 76 ONLINE B/O 10 DLI 10 4 4 2 I/O 8 0 0 8 MOBILE APPL 317 DLI 317 80 158 79 I/O 238 80 0 158 MOBILEM B/O 55 DLI 55 22 22 11 I/O 44 0 0 44 REQUEST APPL 85 DLI 85 55 30 0 I/O 53 11 0 42 STOCK APPL 21 DLI 21 6 10 5 I/O 16 6 0 10 *Total* APPL 575 DLI 575 179 274 122 I/O 421 135 0 286 *Total* B/O 65 DLI 65 26 26 13 I/O 52 0 0 52
Database update activity What databases were updated, and by which program
ISRT, DLET and REPL counts
Application update and back-out
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
11
Avg Avg Tot Tran CPU Process ESAFcall SSID Trancode Count Time Time Count DB2P CHECKING 74 0.038407 0.227590 961 DB2P CREDIT 181 0.016318 0.167907 729 DB2P FINANCE 22 0.027207 0.142876 308 MQS1 FINANCE 22 0.027207 0.142876 308 DB2P MOBILE 56 0.025832 0.192728 282 MQS1 MOBILE 56 0.025832 0.192728 282 DB2P WITHDRAW 21 0.036852 0.211123 273
External subsystem usage According to the IMS log
External subsystem usage by transaction
Only the call counts are available, DB2 and MQ call latencies are not available
DB2 Plan -------- Connection --------- Thread SSID Name Name Type Count DB2P BANKING IMSP IMS MPP PSB=WITHDRAW 21 Class1: Thread Time Avg: Elapsed= .042498 CPU= .035198 Max: Elapsed= .126945 CPU= .107331 Class2: In-DB2 Time Avg: Elapsed= .002525 CPU= .000701 Max: Elapsed= .006999 CPU= .000769 Class3: Suspend Time Avg: Total = .002623 I/O= .000000 Max: Total = .012437 I/O= .000000
According to DB2 accounting
Class 1 (thread) and Class 2 (in-DB2) shows elapsed and CPU times in both IMS and DB2
Suspend analysis can identify the cause
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
12
Shared queues Local-first and false scheduling
Shared queues transaction performance
Avg Avg Avg Avg >1.0 >1.0 Avg Max >0.1
Queue Tran InputQ Process OutputQ Total InputQ Process CPU CPU CPU
IMS ID Type Count Time Time Time Time Time Time Time Time Time
IMS1 GLOBAL 493 0.152755 0.353057 0.111123 0.613036 4.26% 11.76% 0.018737 0.199393 5.77%
IMS1 LOCAL 222 0.164432 0.567554 0.000162 0.732051 4.05% 21.62% 0.028475 0.199393 7.37%
IMS1 LOCALF 8701 0.003339 0.056002 0.208816 0.249786 0.10% 0.92% 0.007174 15.15835 0.32%
IMS2 GLOBAL 472 0.157556 0.432707 0.000675 0.590806 4.45% 15.89% 0.023619 0.236885 6.86%
IMS2 LOCAL 224 0.234427 0.457905 0.000130 0.692367 7.14% 16.52% 0.022103 0.208415 5.58%
IMS2 LOCALF 8131 0.002857 0.065651 0.190159 0.242889 0.00% 0.97% 0.012947 31.13607 0.47%
Total 18243 0.015950 0.089237 0.186943 0.276655 0.42% 2.07% 0.010805 31.13607 0.87%
Region overview MPP ******* DLI Calls ******* WFI **** Schedule Distribution **** Total CPU Sched Trans DB DC Other SUBQ6 Quick =0 =1 >1 ------- ------- ------- ------- ------- ------- ------- ------- ----- ------- ------- ------- Total 181858 4636 16842 178088 41165 24530 41556K 6279 Sched 1408 2131 1097 /Sched 39 1.0 3.6 38.4 8.9 5.3 8963 1.4 Tran 0 2365 14477 /Tran 10 0.3 1.0 10.6 2.4 1.5 2467 Tr/Sch 0 1 13 /Minute 30309 772.7 2807.0 29681.3 6860.8 4088.3 6926016 1046.5 Sc/Min 234.7 355.2 182.8 % Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 %Sched 30.37 45.97 23.66
1. IMS workload broken down by queue type: GLOBAL – shared queues where transactions came in on another system
LOCAL – shared queues where transactions came in on this system
LOCALF – local-first where shared queue was bypassed
2. False schedule and WFI analysis:
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
13
Program Switch: Follow the transaction trail
13
Part 1 of 2
Transaction BANK0001 originated from lterm NEWYORK at 14.58.02.023922
It switched to trancode BANK0010
Trancode BANK0010 then switched to BANK0011
Four (4) further switches to Trancode BANK0015 completed the sequence
The overall transaction response time is always reported against the
originating transaction – 4.467306 seconds – from when the originating
transaction arrived on the message queue to when a response was sent back
to the terminal
The last transaction in the sequence responded back to the terminal – you can
tell because it has an output queue time
Notice the long switch times for the last 2 transactions – this contributed to
almost 0.5 seconds of the response time
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
14
Program Switch: Looking for bottlenecks
14
Part 2 of 2
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
15
DB DB FP FP Program CHKP ID Start time CPU Elapsed Get Upd Get Upd Database AccM NITERUN NITECP1 21.30.23.310 0.2329 1.3150 1332 1 1473 3 CUSTOMER ESDS PAYMENTS DEDB ORDERS CHKP STOCK CHKP WIDGETS CHKP NITERUN NITECP2 21.30.24.657 0.3147 1.4038 1465 2 1521 4 CUSTOMER ESDS PAYMENTS DEDB ORDERS CHKP STOCK CHKP WIDGETS CHKP
BMP analysis For each checkpoint interval, see
Elapsed and CPU times
Database activity
Databases updated and checkpointed
BMP NITERUN issued user checkpoints with an id
Syncpoints were taken at regular intervals, every 1.5 seconds
CPU time was consistent across each of the intervals
Both full function and fast path database activity occurred – gets and updates
One FF and one DEDB database were updated
Three databases were checkpointed – used by the BMP for read-only purposes
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
16 16
Synchronous callout – Tracking Part 2 of 2
BROWSE IPI000.QADATA.JASDEEP2.VB10.IM01001 Record 00000057 More: < >
Command ===> Scroll ===> PAGE
Forwards / Backwards . . HH.MM.SS.THMIJU Time of Day . . HH.MM.SS.THMIJU
Code Description Date 2011-04-20 Wednesday Time (Relative)
/ ---- Search limit reached (+5000) ------------------------- ----------------
01 Input Message TranCode=SMQLG 01.19.00.003853
35 Input Message Enqueue TranCode=SMQLG +0.000001
31 DLI GU TranCode=SMQLG Region=0006 +0.000057
6701 YOUT Sync callout message sent +0.000125
6701 YACK Received ACK +0.000618
6701 YRSP Sync callout response received +0.201386
03 Output Message Response LTerm=OTM00001 +0.201420
35 Output Message Enqueue LTerm=OTM00001 Region=0006 +0.201422
37 Syncpoint Region=0006 +0.201424
37 Syncpoint message transfer Region=0006 +0.201425
33 Free Message +0.201428
5612 End of Phase 2 Syncpoint Program=PSBLG Region=0006 +0.201431
31 Communications GU LTerm=OTM00001 +0.201440
36 Output Message Dequeue LTerm=OTM00001 +0.203894
33 Free Message +0.203895
07 Application Terminate TranCode=SMQLG Region=0006 +0.220357
******************************** Bottom of Data *******************************
Delay due to
sync callout
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
17 17
The sync callout response time is a
large percentage of the overall
processing time: indicative of a delay
Transactions w/ sync-callout
SUMM0001 Data from 10.18.37 19Apr2011 to 10.21.06 19Apr2011
Avg Avg Avg Avg Avg Avg Avg
Tran CPU InputQ Process SyncCout Total SyncCout SyncCout
Trancode Count Time Time Time RespTime IMS Time ACK Cnt NAK Cnt
MYORDER 57760 0.0023 0.0052 0.5673 0.4723 0.5725 1 0
Synchronous callout – Reporting Part 1 of 2
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
19
BROWSE FUW000.QADATA.FBOSP006.SMF.D130530.CICS + Record 00000088 More: < > Command ===> Scroll ===> CSR Navigate < 00.00.01.000000 > Date/Time 2013-05-30 09.22.22.669470 / ———— Tracking ———————————————————————— Thursday 2013-05-30 Time (Relative) CICS Transaction 11.03.41.587515 TranCode=FB66 Program=FBOCCP66 Userid=TWM LTerm=SC0TCP06 Terminal=CP06 RecToken=FUWTCIC/CB6F7079F5587284 Resp=19.441616 CPU=0.016455 DB2=6 ACCT=FTS3.SC0TCP06.6F7079F55872 Task=679
19
Transaction indexes – what are they and is there one for IMS?
Part 1 of 3
IMS Transaction 11.34.22.536129 UTC=11.34.22.536121 TranCode=FBOIAT42 Program=FBOIAP42 Userid=FUNTRM14 LTerm=FUNTRM14 Terminal=SC0TCP14 Region=0002 OrgUOWID=IDDG/CC17F14163997884 SSID=IDDG IMSRel=131 RecToken=IDDG/0000000F00000000 CPU=0.027747 InputQ=0.000356 Process=0.032033 OutputQ=0.000027 TotalTm=0.032416 RegTyp=MPP
DB2 Accounting 11.03.59.237323 TranCode=FB66 Userid=TWM ClientID=FUWTCIC SSID=DB2P ET1=19.43099 ET2=9.486542 CPU1=2.750604 CPU2=2.748489 Suspend=5.217919 ACCT=FTS3.SC0TCP06.6F7079F55872 Source=CICS SEL=3 CAL=3 GetPage=43839 MaxLock=1 LUWID=FTS3/DBA6LU/CB6F7079F61B/0002
CICS has CMF written to SMF 110
DB2 has accounting written to SMF 101
You can create an IMS accounting record from the IMS log
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
20
Code... 56FA Transaction Statistics STCK... CBFC740BD90293A4 LSN.... 000000001C254F89 Date... 2014-06-23 Monday Time... 14.58.01.912361.227 JOBNAME.... 'IMSPMSG1' PROGRAM.... 'BANKING' TRAN....... 'BANK2' CPU........ 0.009593 DBIO....... 0.068882 LockWait... 0.000544 IOcalls.... +15 VSAMread... +7 VSAMwrite.. +2 OSAMread... +6 OSAMwrite.. +1 ESAFcalls.. +12
Benefits of Collecting 56FA – transaction accounting statistics
TRANSTAT=YES cuts type x'56FA' transaction accounting statistics
records at transaction-end and BMP-checkpoint.
Overcomes the limitations of type x'07' which is cut at program
schedule termination and often accumulates statistics for many
transactions.
Without 56FA you cannot get accurate transaction CPU time or other
useful transaction-level details.
You
need
the
56FA
Part 2 of 3
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
21
IMS transaction index
21
IMSPerformanceAnalyzer
IMSProblemInvestigator
IMS PA batch job condensesall IMS log records forone IMS transaction intoone Transaction Index record
Save processing time by usingTransaction Index as inputto IMS PA instead of IMS log
Transaction Index contains performanceinformation (such as response time) derivedfrom IMS log fields, enabling more intelligentfiltering and analysis in IMS PI
IMS log
TransactionIndex
Whether Full or Exception, all the accumulated information from the IMS log about each transaction is condensed into a single transaction index record.
Use the transaction index as input into:
IMS PA reporting - Instead of re-processing large SLDS log files
IMS PI analysis - To enable more intelligent problem detection
ETL into Big Data “Inflight” option allows
for continuous log
processing without
missing a beat.
Part 3 of 3
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
22
How to address IMS Log privacy issues using IMS Problem Investigator
IMS SLDS IMS PI SCRUB
(Batch) Scrubbed
extract
IMSPI only feature:
IMS PI ISPF dialog:
Interactive analysis
IMS PI Batch:
Reporting, extract and CSV
IMS PI SCRUB
(Exit)
1
2
1 SCRUB utility 1. Systems Programmer creates a copy of the original SLDS with confidential data scrubbed
2. The scrubbed extract data set is made available to all IMS PI users – data set to <<CLEARED>>
3. 100% secure – user has no access to confidential data
SCRUB exit (ALZUAUTH) – IMSPI only 1. All IMS PI users have access to the original (un-scrubbed) SLDS
2. All IMS PI ISPF dialog and batch services scrub the log records prior to display or reporting
3. Not secure – other utilities such as ISPF browse or DFSERA10 will display the confidential data
2
External
Problem
Analysis
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
23
IMS Problem Investigator batch reporting
Code Description Date 2011-06-17 Friday Time Elapsed LSN Rec#
---- ------------------------------------------------------ ------------------- --------------- ---------------- ---------
01 Input Message 21.57.00.950009.909 0.078041 0000000005AD82B0 9752
TranCode=CDB41PDS Userid=EVHWBUDR LTerm=4012 Terminal=4012
OrgUOWID=BUDR/C7EF90B3BC1BD08C Port=4012 LogToken=C7EE6AA5E342DC08
SSN=087124 Socket=PERS CM=1 SL=0 Source=Connect
35 Input Message Enqueue 21.57.00.950025.284 0.000015 0000000005AD82B1 9753
TranCode=CDB41PDS Userid=EVHWBUDR LTerm=4012 Terminal=4012
OrgUOWID=BUDR/C7EF90B3BC1BD08C Port=4012 LogToken=C7EE6AA5E342DC08
SSN=087124 Socket=PERS CM=1 SL=0
08 Application Start 21.57.00.950310.565 0.000285 0000000005AD82B2 9754
TranCode=CDB41PDS Region=000F RecToken=BUDR/0021DB9C00000000
RegTyp=MPP TClass=2D TPrty=01
5607 Start of UOR 21.57.00.950310.627 0.000000 0000000005AD82B3 9755
Program=CDB41PT Region=000F IMSID=BUDR RecToken=BUDR/0021DB9C00000000
31 DLI GU 21.57.00.950329.127 0.000018 0000000005AD82B4 9756
TranCode=CDB41PDS Region=000F OrgUOWID=BUDR/C7EF90B3BC1BD08C
RecToken=BUDR/0021DB9C00000000
5600 Sign-on to ESAF 21.57.01.048592.688 0.098263 0000000005AD82B5 9757
Region=000F IMSID=BUDR RecToken=BUDR/0021DB9C00000000 SSID=DSBR
SHORT batch report mode – similar to the ISPF dialog view
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
24
Want to rid your IMS GEN of no longer used resources?
Uses IMS checkpoint records:
4004=Transactions, 4006=Databases, 4007=Programs
For databases:
Start of checkpoint interval Date=2014-06-26 Thursday Time=17.51.31
DDIR DDIROPEN DDIRDA CHKWOPN DDIRGOPN Used? -------- -------- ------ ------- -------- ----- ACCOUNTS 40 20 00 00 YES BANKING 00 00 00 00 NO ORDERS 00 00 00 00 NO STOCK 40 20 00 00 YES
DDIR Database name
DDIROPEN At least one DCB is opened
DDIRDA DMB has been dynamically allocated
CHKWOPN DB was open
DDIRGOPN GBLSTATUS = OPEN
Used? YES – at least one of the above is set, DB is in use or has been used
NO – none of the above are set, DB has not been used
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
25
IMS Problem Investigator - IMS Trace Record Analysis /DLI Trace
DLI Trace events in IMS log:
67FF SNAP Trace when transaction abends
67FA /TRACE SET ON TABLE DL/I , LOCK etc
See every DLI call,
then associate with
the database updates
(50’s)
IMS Trace Table records are now interpreted, and can be tracked against the problem transaction:
Part 1 of 2
TX
Transaction Index
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
26
IMS Problem Investigator - IMS Trace Record Analysis /TRACE SET ON TABLE LOCK
Locking events, including
SUSPEND and RESUME
and any Associated Delays
• Dispatcher
• DLI and Lock
• Log Router
• Scheduler
• Queue Manager
• Latch
• LU 6.2 (APPC)
• Fast Path
• RRS
• DASD log
• External Subsystem
• OTMA
• Storage Manager
Part 2 of 2
{{{
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
27
CICS DBCTL
For all CICS transaction you can analyze CMF (SMF 110) records.
However CMF records do not contain all of the details of the IMS events
that are triggered by DL/I requests. Those details are in the IMS log.
Transaction Analysis Workbench enables you to analyze CICS DBCTL
transactions using:
–CMF records
– IMS log records
–A combined view of both
In particular Workbench enables you to analyze exceptions:
CICS DBCTL transactions that abended or had a long response time.
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
28
CICS-DBCTL summary – from both their perspectives
ss
CICS
ss
IMS
2014-03-26 Wednesday CICS-DBCTL Summary Page 1 Tran APPLID CMF Count Response CPU Time IMS Reqs IMS Wait ABEND Rate/Sec ---- -------- --------- -------- -------- -------- -------- -------- -------- BANK CICSP1 60 11.12982 0.008967 35 4.256977 10 0 08 Count Elapsed CPU Time StaDelay Schedule IC Wait PS Wait --------- -------- -------- -------- -------- -------- -------- 42 10.94999 0.004092 0.011668 0.000183 0 0 07 Count DB call DB Gets DB Upds IO Count IO Time LockWait --------- -------- -------- -------- -------- -------- -------- 41 33 13 19 4 0.003438 3.980170 FP Count FP call FP Gets FP Upds FP Wait FP Fail --------- -------- -------- -------- -------- -------- 41 19 7 11 0 7 Synctime Phase 1 Phase 2 FP PH2 OTHREAD -------- -------- -------- -------- -------- 0.011938 0.006555 0.005383 0.002232 0.017659
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
30
IMS-DB2 transaction – Tracking end-to-end Part 1 of 4
BROWSE FUW000.QADATA.FBOSPM3C.IMS.D131011.INDEX Record 00006876 More: < > Command ===> Scroll ===> CSR Navigate < 00.00.01.000000 > Date/Time 2013-10-11 11.29.17.925791 / ———— Tracking ———————————————————————— Friday 2013-10-11 Time (LOCAL) CA01 IMS Transaction 11.33.58.759345 TranCode=FBOIAT42 Program=FBOIAP42 Userid=FUNTRM14 LTerm=FUNTRM14 Terminal=SC0TCP14 SSID=IDDG Region=0001 RecToken=IDDG/0000000E00000000 OrgUOWID=IDDG/CC17F12AB6B85204 InputQ=0.000332 Process=1.094017 OutputQ=0.000028 TotalTm=0.134377 CPU=0.111167 ———————————————————————————————————————————————————————————————————————————— . . . ———————————————————————————————————————————————————————————————————————————— 003 DB2 Thread accounting DBA6 11.33.58.893342 TranCode=FBOIAP42 Program=FBOIAP42 Userid=FUNTRM14 RecToken=IDDG/0000000E00000000 ClientID=IDDG Region=0001 LUWID=FTS3/DBA6LU/CC17F12AB853/0002 RESP=1.086945 CPU1=0.107330 CPU2=0.000688 I/O3=0 Source=IMS_MPP
SEL=4 DEL=1 CAL=1 LogRecs=9777 GetPage=19777 UpdPage=9776 MaxLock=162 ———————————————————————————————————————————————————————————————————————————— ******************************* Bottom of Data ********************************
1. Both the IMS and DB2 index records show there is a problem
2. The DB2 accounting (IFCID 003) shows excessive logging, buffering and locking – where and why?
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
31
IMS-DB2 transaction – Tracking end-to-end Part 2 of 4
BROWSE FUW000.QADATA.FBOSPM4C.IMS.D130703.INDEX Record 00005529 More: < > Command ===> Scroll ===> CSR Navigate < 00.00.01.000000 > Date/Time 2013-07-03 16.33.53.849552 / ———— Tracking ———————————————————————— Wednesday 2013-07-03 Time (Relative) CA01 IMS Transaction TranCode=FBOICT42 Region=0002 16.40.55.075503 01 Input Message TranCode=FBOICT42 +0.000000 31 DLI GU TranCode=FBOICT42 Region=0002 +0.000266 5600 Sign-on to ESAF Region=0002 SSID=DBA6 +0.005517 112 Thread allocate FBOIAP42 DBA6 +0.005996 177 Package allocation FBOIAP42 DBA6 +0.006304 380 SP entry FBOSPM4C STMT=001044 DBA6 +0.006511 177 Package allocation FBOSPM4C DBA6 +0.006681 060 SQL SELECT STMT=000009 DBA6 +0.006796 058 SQL call completion SQLCODE=0 STMT=000009 DBA6 +0.016090 061 SQL DELETE STMT=000011 DBA6 +0.016387 0020 Begin UR +0.017584 325 Trigger entry USERDEL STMT=000011 DBA6 +0.017762 177 Package allocation USERDEL DBA6 +0.017836 061 SQL INSERT STMT=000002 DBA6 +0.017887 058 SQL call completion SQLCODE=0 STMT=000002 DBA6 +0.018139 Continued on next foil...
1. IMS transaction starts with an Index record (CA01)
2. IMS log (green), DB2 trace (yellow) and DB2 log (turquoise) events are shown
3. Transaction calls a DB2 stored procedure which issues SQL calls, call results are shown
IMS
DB2
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
32
IMS-DB2 transaction – Tracking end-to-end Part 3 of 4
... Continuation from previous foil 061 SQL DELETE STMT=000003 DBA6 +0.018179 0600 Delete from a data page . . . +0.019696 0600 Delete from a data page +1.082712 058 SQL call completion SQLCODE=0 STMT=000003 DBA6 +1.083240 325 Trigger exit SQLCODE=0 DBA6 +1.083254 058 SQL call completion SQLCODE=0 STMT=000011 DBA6 +1.083310 499 SP statement execution detail DBA6 +1.083370 380 SP exit FBOSPM4C SQLCODE=0 STMT=001044 DBA6 +1.083400 053 SQL request SQLCODE=0 STMT=001044 DBA6 +1.083436 5600 Commit Prepare starting Region=0002 SSID=DBA6 +1.084442 084 Prepare start DBA6 +1.084894 0020 End commit phase 1 +1.085008 085 Prepare end DBA6 +1.086509 03 Output Message Response LTerm=FUNTRM07 +1.086577 3730 Syncpoint End of Phase 1 Region=0002 +1.086598 074 Terminate thread start DBA6 +1.086969 0020 Begin commit phase 2 +1.087056 0020 End commit phase 2 +1.087728 003 Thread accounting DBA6 +1.089002 075 Terminate thread end DBA6 +1.089944 5600 Commit Continue completed Region=0002 SSID=DBA6 +1.089970 37 Syncpoint Message Transfer Region=0002 +1.089987 5612 Syncpoint End of Phase 2 Program=FBOIAP42 Region=0002 +1.090012 31 Communications GU LTerm=FUNTRM07 +1.090054 36 Output Message Dequeue LTerm=FUNTRM07 +1.093952 ******************************* Bottom of Data ********************************
IMS
SYNCPOINT
DB2
Delay
4. SQL DELETE causes trigger to delete many more table rows, causing the delay
© 2015 IBM Corporation
IMS Technical Conference Australia 2015
33
+0120 QW0058ID... Scan information +0120 Scan type.... 'SEQD' Rows processed... +24069 +0130 Rows examined.... +24069 +0138 Rows qualified after stage 1... +24069 +0140 Rows qualified after stage 2... +1 +0148 Rows inserted.... +0 +0150 Rows updated... +0 +0158 Rows deleted... +0 +0160 Pages scanned.... +428 +0164 Pages scanned (RI)... +0 +0168 Rows deleted (RI).... +0 +0170 Pages scanned (LOB).... +0 +0174 Pages updated (LOB).... +0 +01A0 QW0058TY... Statement-level information +01A0 SQL statement type... 4000 +01A8 Statement ID... +28917 +01B0 Sync reads... +0 Getpages... +428 +01C0 Rows examined.... +24069 +01C8 Rows processed... +0 Sorts...... +0 +01D8 Index scans.... +0 +01E0 Table space scans.... +1 +01E8 Buffer writes.... +0 +01F0 Parallel groups.... +0 +01F8 In-DB2 elapsed... 0.008537
Enhanced support for DB2 trace (IFCID) records Part 4 of 4
Detailed formatting of IFCID-specific fields:
+0120 QW0058ID... 'SEQD' Scan type Off QW0058IX... 'INDX' Index On QW0058SD... 'SEQD' Sequential data Off QW0058SW... 'SEQW' Sequential data workfile