View previous topic :: View next topic
|
Author |
Message |
Ganesh Kalam Warnings : 1 New User
Joined: 11 Aug 2005 Posts: 45 Location: India
|
|
|
|
I need to know what steps need to be taken while creating DASD or TAPE datasets, to avoid permanent retention of files and also to achieve optimum performance.
For eg.. BLKSIZE=0
EXPDT=98000
If you have any tips like these, please post. |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19244 Location: Inside the Matrix
|
|
|
|
Hello,
You need to talk with the storage administrators for your system.
They can tell you what preferred on your system for all of your storage allocations. |
|
Back to top |
|
|
Anuj Dhawan
Superior Member
Joined: 22 Apr 2006 Posts: 6250 Location: Mumbai, India
|
|
|
|
Hi,
I think every site has it's own MANAGEMENT class specified in there system arcticture. None of the files, DASD or TAPE, is given, by default, a permanent retention unless specified in respective JCLs. However, Storage guys are the best to answer as suggested.
There are no direct tip as such for this. |
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
Yes, it is all site specific, depending on how the storage group have set things up.
One little tip, BLKSIZE=0 in certain circumstances causes far more problems than it solves. Code only RECFM & LRECL to avaoid these problems for DASD datasets. |
|
Back to top |
|
|
Anuj Dhawan
Superior Member
Joined: 22 Apr 2006 Posts: 6250 Location: Mumbai, India
|
|
|
|
Hi expat,
I thought that BLKSIZE=0 actually saves the DASD sapce. if specified, DSN will have BLKSIZE=0 till the DSN is not written to, once respective processing is done with DSN, i.e. written to, it's BLKSIZE is whatever decided by the System (by SMS, right?). If so then I'm little unclear on Your this comment:
Quote: |
in certain circumstances causes far more problems than it solves |
Please assist. |
|
Back to top |
|
|
CICS Guy
Senior Member
Joined: 18 Jul 2007 Posts: 2146 Location: At my coffee table
|
|
|
|
expat wrote: |
One little tip, BLKSIZE=0 in certain circumstances causes far more problems than it solves. |
Yea, expat, please expand on this..... |
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
BLKSIZE=0 will create a dataset with exactly that, usually when the dataset is created but never opened for output. By only coding the RECFM and LRECL parameters the SDB routines detect no BLKSIZE specified and are invoked to create the dataset with a valid blksize.
HSM does not process datasets with BLKSIZE=0, and given that around 30% of developers are unaware of the RLSE parameter in JCL, they just use up space for no good reason and then come whinging to me when they run out. |
|
Back to top |
|
|
Anuj Dhawan
Superior Member
Joined: 22 Apr 2006 Posts: 6250 Location: Mumbai, India
|
|
|
|
Hi,
Coming back to this thread after a while, but you (Expat) are still here, so I can ask some more dumb questions..I know..
expat wrote: |
HSM does not process datasets with BLKSIZE=0 |
is it good or bad ? If HSM is not going to process them - does that mean 'am actually saving some CPU seconds - in this manner is it good or bad ?
Quote: |
and given that around 30% of developers are unaware of the RLSE parameter in JCL, they just use up space for no good reason and then come whinging to me when they run out. |
So does that mean using RLSE with BLKSIZE=0 is rather appropriate ? If BLKSIZE is already coded to zero, how the "they were run out" of sapce, please assist.
I might run some test for this but there are two reasons why I can't do that - my Mainframe is not "up" and even if I run some tests I don't know where exactly to look at to know if "that combination" had some effect ... |
|
Back to top |
|
|
expat
Global Moderator
Joined: 14 Mar 2007 Posts: 8797 Location: Welsh Wales
|
|
|
|
Quote: |
is it good or bad ? If HSM is not going to process them - does that mean 'am actually saving some CPU seconds - in this manner is it good or bad ? |
HSM will attempt to process the datasets and complain that the dataset has an invalid blocksize, and write an error message in the backup / migrate logs.
Quote: |
So does that mean using RLSE with BLKSIZE=0 is rather appropriate ? If BLKSIZE is already coded to zero, how the "they were run out" of sapce, please assist. |
RLSE and BLKSIZE are independant. My comment was that a lot of developers do not use RLSE in their JCL and allocate space and not use it. I usually set up the MGMTCLAS to release unused space for , so no longer a problem for me.
So, testing combinations of RLSE and BLKSIZE=0 should not yield any useful information. |
|
Back to top |
|
|
|