Page 1 of 1

Performance is my question????

PostPosted: Tue Jan 18, 2011 7:09 am
by fornanthakumar
Hi,

Scenario : 1. Creating one dataset 2. Processing by the JOB 3. Used in successor JOB 4. Delete the Dataset.

My question here is,

Instead of having IEFBR14 step in successor JOB to delete the dataset, if i modify the creation IEFBR14 step in first JOB like below
//CREATE EXEC PGM=IEFBR14
//CREATE1 DD DSN=CHN0.SDSF.APPS123,DISP=(MOD,CATLG,DELETE),
// SPACE=(CYL,(1,1),RLSE),VOL=SER=SBD138,
// DCB=(LRECL=80,BLKSIZE=8000,RECFM=FB),
// UNIT=SYSDA

It will create the dataset even though the dataset exist or not. Will it give low performance.

Need your input... :)

Re: Performance is my question????

PostPosted: Tue Jan 18, 2011 7:39 am
by steve-myers
Putting the volume serial into the allocation raises a lot of questions. I would not do it. Something like this is usually done in two steps:

//A EXEC PGM=IEFBR14
//ADS DD DISP=(MOD,DELETE),UNIT=SYSALLDA,SPACE=(TRK,0),
// DSN=whatever
//B EXEC PGM=IEFBR14
//ADS DD DISP=(NEW,CATLG),...,DSN=*.A.ADS

The first step will delete the dataset, if it exists. The second step will allocate a new instance of the dataset. This will not work if the dataset exists on tape.

I can't think of any unusual performance issues.

Re: Performance is my question????

PostPosted: Tue Jan 18, 2011 9:14 am
by dick scherrer
Hello,

It will create the dataset even though the dataset exist or not.
No, it will not. . . If the dataset exists, it will not be created again.

Depending on how your system is managed issuing a CATLG for an existing dataset may cause unacceptable results.

As mentioned, you should not mention a specific volser in your allocation jcl unless specifically directed to do so. You should use the system standards or ask a senior or the storage management people how to allocate your data.

Re: Performance is my question????

PostPosted: Tue Jan 18, 2011 11:03 am
by NicC
And you do not need to code DCB=... just code
LRECL=80,BLKSIZE=0,RECFM=FB

Note BLKSIZE=0 allowing the system to decide the most appropriate blocksize

Re: Performance is my question????

PostPosted: Tue Jan 18, 2011 12:18 pm
by steve-myers
NicC wrote:And you do not need to code DCB=... just code
LRECL=80,BLKSIZE=0,RECFM=FB

Note BLKSIZE=0 allowing the system to decide the most appropriate blocksize
NicC is correct, but I learned DCB=(...) 40+ years ago, and can't break the habit! No need for a newbie to get in the habit! Heavens! I'm making it sound like quitting smoking!

I think I'd write RECFM=FB,LRECL=80,DSORG=PS unless I'm allocating a PDS.

Re: Performance is my question????

PostPosted: Tue Jan 18, 2011 12:42 pm
by stevexff
IEFBR14 does what it says; BR 14. Whatever you do to it, you won't have a performance problem. But if the dataset you reference has been migrated by HSM, it may take a little time to recall it before you delete it...