To reduce the Batch job run time

harkar · New User Joined: 11 Nov 2006 Posts: 7 Location: Chennai

Hi ,

Can some one suggest me how to reduce the batch job run time?

Regards
Chandu

stodolas · Posted: Mon Nov 19, 2007 9:21 pm

Remove some steps, get a bigger mainframe, use empty input files.

dbzTHEdinosauer · Posted: Mon Nov 19, 2007 9:53 pm

harkar,

your question provided so little information, was so vague, displayed so little understanding of computers

that

not only did you deserve Steve's answer, I wish I had been clever enough to say it first.

yeah for Steve, very nice way of handling such a ridiculous question.

dick scherrer · Posted: Mon Nov 19, 2007 11:30 pm

Hello Chandu,

Now that you have been somewhat "beaten up", let me suggest that to get anything useful from the forum, you have to provide something for us to work with - you did not.

If you will clarify what run time you want to reduce, we may be able to offer suggestions. You need to describe the process, what the volume of the input and output is, how long the process takes, and what you consider acceptable run time.

To go along with what others have already posted, the way your question is presented, the only answer would be some form of "reduce the resources used" or "find somewhere faster to run it".

Bitneuker · Posted: Tue Nov 20, 2007 3:04 am

Another answer might be........design your applications the way you should.

harkar · New User Joined: 11 Nov 2006 Posts: 7 Location: Chennai

I mean to say,let us assume we have a COBOL DB2 program and it is taking most of the CPU time, how to reduce the CPU time? like tuning the DB2 query without impacting the functionality , like wise do you have any other suggestions where the we can implement few things to reduce the CPU time??

Please excuse me if my previous questions was not clear!!!

Chandu

dick scherrer · Posted: Fri Dec 12, 2008 10:24 am

Welcome back to your topic

Only been a year

As before, you haven't provided anything for us to use to help you

Before you can reduce the cpu time, you need to identify which part(s) of the main problem program are using the most time.

You need to consider how many rows are processed by the code and then determine if there might be a better way to implement that code.

Suggest you do some analysis as to how much work is actually being done (much of it may just be overhead) and we can go from there.

You might also talk with your dbas as they have tools to monitor queries and should be able to spot problem sql.

Anuj Dhawan · Posted: Fri Dec 12, 2008 12:29 pm

Rajeev Saluja · New User Joined: 28 Mar 2008 Posts: 6

Hi,

In regard your query..There is a suggestion

- Relocate the steps in JCL (e.g Backup File steps) which can be run parallel in some other JCL so as to save execution time.

- Can remove the wait time from the JCL (if it been used)

-Can alter the JCL to run in 2 phases (phases-1, phases-2) which can also run parallely but not impacting/editing the data in the file due to the change.

Terry Heinze · Posted: Sat Dec 13, 2008 1:14 am

After all of the above, consider using batch pipes.

MBabu · Posted: Sat Dec 13, 2008 1:51 am

The question, such as it was, was about CPU time, not elapsed time. Parallel execution will not affect CPU time and might even increase it very slightly. But as others have said, there is absolutely no information to work with in the question, so any answer is just speculation. All we know is that there is some sort of COBOL program and that DB2 is somehow involved. Sort of like asking "I have this thing and I want it to be different. Can anyone help?"

Rajeev Saluja · New User Joined: 28 Mar 2008 Posts: 6

Dears,

I am not sure if you consider the CPU usage might be not full while executing a particular job/set of jobs. If CPU can accomodate the same space can be utilized in earlier scenario thereby reducing the load on existing job. If in a small example consider your job has BACKUP file step. The same step if you take earlier during backup in previous job steps it wont change/impact much if we introduce the backup step in the same earlier job.

dick scherrer · Posted: Tue Dec 16, 2008 8:11 pm

More efficient block sizes, faster media, and proper scheduling are generic ways to improve sort performance.

enrico-sorichetti · Posted: Tue Dec 16, 2008 8:12 pm

All the suggestions given are part of any decent organization standards and practices manual
under the headings
... <general procedures layout> or something similar
... <step sequencing> for backup archivals restarts
... <resource availability checking> (*)

unless the OP has given evidence of unreasonable elapsed/run times
there very little we can do for him

(*) with automated tape libraries now is not an issue any longer
but once upon a time with manual tape handling a proper tape volume pull list and preparation was crucial to a smooth job flow