Page 1 of 2

STROBE

PostPosted: Sun Jan 20, 2013 7:48 pm
by vikkysharma
i am looking for STROBE tool manual or any study material. anybody can help me for the same.

Re: STROBE

PostPosted: Sun Jan 20, 2013 8:10 pm
by Robert Sample
The vendor for STROBE makes the manuals available on their web site. If your site has a license for STROBE, you can get a sign on from the vendor and download the manuals yourself. If your site does not have a license for STROBE, it is illegal to post the manuals or links to them as they are under copyright.

Re: STROBE

PostPosted: Sun Jan 20, 2013 8:33 pm
by enrico-sorichetti
and ...
what does a request for manuals have to do with Suggestions & Feedback ??? :twisted:

Re: STROBE

PostPosted: Mon Jan 21, 2013 6:46 pm
by vikkysharma
Thansk for your suggestions Robert and Enrico, i am new for strobe and want to optimize a cobol program.
trru my stroobe report i found that INSPECT verb is taking around 80% CPU utilization. i am trying to optimize this with replacement of PERFORM loop.
please suggest if you have any more Ideas to optimization.

Re: STROBE

PostPosted: Mon Jan 21, 2013 7:00 pm
by BillyBoyo
We can move the topic to the Cobol part of the forum then.

Please post the code that you regard as problematic, including all data definitions, in the Code tags please.

Re: STROBE

PostPosted: Tue Jan 22, 2013 9:23 pm
by dick scherrer
Hello and welcome to the forum,

While INSPECT has a "heavy" resource requirement, make sure this is actually what is causing the poor performance.

How is "poor performance" defined for this process? Is the process running too long? While the INSPECT is quite visable, it may not be a problem at all.

Re: STROBE

PostPosted: Wed Jan 23, 2013 8:06 pm
by Ed Goodman
In general, work the STROBE report from top to bottom. First, look at the profile (near the top) and see what percentages are in CPU/WAIT/QUEUED. Most likely, you have a high WAIT percentage. If so, then drop down to the WAIT sub reports and see where it's coming from.

If you really do have a high CPU percentage, then you are doing the right thing by chasing down the verb(s) using a lot of CPU.

The hardest thing to get used to with STROBE and APA is that you are dealing with percentages. You may be seeing a verb using 85% of CPU, but out of how much time? If it's 85% of two hours, then you can probably make a difference. If it's 85% of 1 second, you may be past the point of diminishing returns.

The most important thing is to test before and after. Sure, you can change that one verb to not have 85%, but did the total go down after you were done?

Re: STROBE

PostPosted: Mon Jan 28, 2013 8:11 pm
by vikkysharma
Hi all ,
i am very new to STROBE, i have attach two screen shot of Strobe report, anybody can lillte explain me how i can strat my analsys to tune up my job.

MEASUREMENT SESSION DATA  . . . . . . . . . . . . . . . . . . . . . . . . . #MSD
TIME DISTRIBUTION OF ACTIVITY LEVEL . . . . . . . . . . . . . . . . . . . . #TDA
RESOURCE DEMAND DISTRIBUTION  . . . . . . . . . . . . . . . . . . . . . . . #RDD
WORKING SET SIZE THROUGH TIME . . . . . . . . . . . . . . . . . . . . . . . #WSS
WAIT TIME BY MODULE . . . . . . . . . . . . . . . . . . . . . . . . . . . . #WTM
DATA SET CHARACTERISTICS  . . . . . . . . . . . . . . . . . . . . . . . . . #DSC
DATA SET CHARACTERISTICS SUPPLEMENT . . . . . . . . . . . . . . . . . . . . #DSS
I/O FACILITY UTILIZATION SUMMARY  . . . . . . . . . . . . . . . . . . . . . #IOF
MOST INTENSIVELY EXECUTED PROCEDURES  . . . . . . . . . . . . . . . . . . . #IEP
PROGRAM SECTION USAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . #PSU
PROGRAM USAGE BY PROCEDURE  . . . . . . . . . . . . . . . . . . . . . . . . #PUP
DASD USAGE BY CYLINDER  . . . . . . . . . . . . . . . . . . . . . . . . . . #DUC
ATTRIBUTION OF CPU EXECUTION TIME . . . . . . . . . . . . . . . . . . . . . #ACE
ATTRIBUTION OF CPU WAIT TIME  . . . . . . . . . . . . . . . . . . . . . . . #ACW


and second screen is :

#IEP                   ** MOST INTENSIVELY EXECUTED PROCEDURES **                 
                                                                               
                                                                               
MODULE   SECTION   LINE     PROCEDURE                         START   % CPU TIME
 NAME     NAME    NUMBER       NAME                            LOC    SOLO TOTAL
                                                                               
.VSAM    IDA019L1                  VSAM RECORD MANAGEMENT            94.54 95.36
GTPGM1   GTPGM1                                              002DC0   1.22  1.22
.COBLIB  IGZCPAC                   COBPACK  GENERAL                    .69   .72
GTPGM1   GTPGM1                                              001A00    .47   .53
GTPGM1   GTPGM1                                              001A40    .31   .31
GTPGM1   GTPGM1                                              003140    .22   .25
.IOCS    IGG019AJ                  QSAM SMPL PUT LOCATE VAR            .16   .19
.COBLIB  IGZCPCO                   PARTITION COMMUNICATION             .19   .19
.IOCS    IGG019BK                  DATA MANAGEMENT SERVICES            .03   .13
.COBLIB  IGZEQBL                   QSAM INIT I/O  & EXITS              .09   .09


(coded)

Re: STROBE

PostPosted: Mon Jan 28, 2013 9:19 pm
by Robert Sample
You need to learn how to use the Code button on this forum to preserve spacing of mainframe reports.

Furthermore, you didn't provide some of the key metrics required -- such as number of samples taken and total CPU time used. However, it appears that VSAM record management is using most of the CPU time in your job -- which means you need to investigate whether the program is doing direct or sequential reads, uses NSR / LSR appropriately, has appropriate buffers (both BUFND and BUFNI), and is this a different program than you had the INSPECT issue with?

Re: STROBE

PostPosted: Tue Jan 29, 2013 12:37 am
by vikkysharma
yes Robert this is different program , in that program i replace the inspect with Perform and refrence- modification it ran fine and save almost 50% CPU.