Portal | Manuals | References | Downloads | Info | Programs | JCLs | Master the Mainframes
IBM Mainframe Computers Forums Index
 
Register
 
IBM Mainframe Computers Forums Index Mainframe: Search IBM Mainframe Forum: FAQ Memberlist Usergroups Profile Log in to check your private messages Log in
 

 

Sort pulling down the performance

 
Post new topic   Reply to topic    IBMMAINFRAMES.com Support Forums -> DFSORT/ICETOOL
View previous topic :: :: View next topic  
Author Message
rathin

New User


Joined: 25 Sep 2008
Posts: 2
Location: chennai

PostPosted: Wed Nov 05, 2008 2:28 pm    Post subject: Sort pulling down the performance
Reply with quote

this is my sort cntrl card.

Code:

   SORT FIELDS=(37,05,CH,A)           
   INCLUDE COND=(902,1,CH,EQ,C'B',OR,
               902,1,CH,EQ,C'P',OR,
               902,1,CH,EQ,C'I')


the input is a tape file . this step ( sorting and filtering ) alone runs for 30 - 35 mts. i need to save time on this..

what can be done .. please provide me and alternative sol..


i splitted the cntrl card to two . but no use ..
can ICETOOL help me for this.. ??

please respond[/b]
Back to top
View user's profile Send private message

dick scherrer

Site Director


Joined: 23 Nov 2006
Posts: 19270
Location: Inside the Matrix

PostPosted: Wed Nov 05, 2008 2:30 pm    Post subject:
Reply with quote

Hello,

How many total records are on the tape? How many volumes does the input use?

How much output data is written?
Back to top
View user's profile Send private message
expat

Global Moderator


Joined: 14 Mar 2007
Posts: 8593
Location: Back in jolly old England

PostPosted: Wed Nov 05, 2008 2:38 pm    Post subject:
Reply with quote

Where is the JCL for the sort step ?
Where is the SYSOUT from the sort step ?

What is the DP of the job ?
- Is it slow batch, normal batch, hot batch. (Terms in use at my current site - but have been used elsewhere). Something you will need to check with your performance & capacity people, or systems programmers if the yare responsible for this.

What else is running against the job ? - Other processes may be running against your job slowing it down, especially if they have a higher DP than your job.
Back to top
View user's profile Send private message
Frank Yaeger

DFSORT Moderator


Joined: 15 Feb 2005
Posts: 7130
Location: San Jose, CA

PostPosted: Wed Nov 05, 2008 11:03 pm    Post subject:
Reply with quote

rathin,

The DFSORT control statements are fine as far as performance goes. You're removing records before sorting them, which helps performance.

It may be that the amount of data requires that much processing time.

However, you might be able to make some changes that will help performance such as using optimum BLKSIZEs for the data sets, or tuning the sort in some way.

If you'd like me to have our performance expert take a look, add the following to your job to get diagnostic messages:

//SORTDIAG DD DUMMY

Then rerun it and send me the complete JES and DFSORT messages offline (yaeger@us.ibm.com). Put "DFSORT" in your Subject line to catch my attention.
Back to top
View user's profile Send private message
Dave Betten

New User


Joined: 24 Jan 2006
Posts: 26

PostPosted: Thu Nov 13, 2008 1:15 am    Post subject:
Reply with quote

Frank passed along your ouptuts so I could take a look at them. From a DFSORT tuning perspective, everything looks fine. You read in about 17.8 million records but only about 40 thousand are actually selected. So not much data is actually sorted and DFSORT was able to use Hiperspace for all of the intermediate storage. The input to the sort is coming from two tape data sets and I suspect the elapsed time is being gated by the speed of the tape drives. If the number of records selected is normally such a small percentage of the file, you might consider splitting this job as follows"

Job 1 reads tape input 1 using SORT FIELDS=COPY with your existing INCLUDE statement.

Job 2 runs concurrent to JOB 1 reading tape input 2. This also uses SORT FIELDS=COPY with your existing INCLUDE statement.

Job 3 sorts the small output files from Jobs 1 and 2 with your original sort fields - SORT FIELDS=(37,05,CH,A). This job can begin as soon as both JOB 1 and JOB 2 are complete.

This at least gets the reading of the two tape data sets to happen in parallel. We could have Jobs 1 and 2 sort the records and have Job 3 do a merge but since so few records are selected, I prefer Job 3 do the sort.
Back to top
View user's profile Send private message
View previous topic :: :: View next topic  
Post new topic   Reply to topic    IBMMAINFRAMES.com Support Forums -> DFSORT/ICETOOL All times are GMT + 6 Hours
Page 1 of 1

 

Search our Forum:

Similar Topics
Topic Author Forum Replies Posted
No new posts How to change 'K' or 'M' use Sort vice_versa DFSORT/ICETOOL 5 Thu May 18, 2017 7:11 am
No new posts Adding big TEXT lines to each record ... bshkris SYNCSORT 4 Sat May 06, 2017 1:40 am
This topic is locked: you cannot edit posts or make replies. SORT trick needed bshkris SYNCSORT 6 Tue May 02, 2017 4:35 am
No new posts SORT JSON type of data maxsubrat DFSORT/ICETOOL 8 Wed Apr 19, 2017 6:01 pm
No new posts Sort Large record length cmsmoon DFSORT/ICETOOL 14 Tue Apr 11, 2017 5:49 pm


Facebook
Back to Top
 
Mainframe Wiki | Forum Rules | Bookmarks | Subscriptions | FAQ | Tutorials | Contact Us