how to allocate very big dataset

JES, JES2, JCL utilities, IDCAMS, Compile & Run JCLs, PROCs etc...

how to allocate very big dataset

Postby samb01 » Fri Dec 16, 2011 8:17 pm


i a have a big dataset et i would like to know what is the best kind to allocate the space ?

I don't really know which is the size the final dataset... I juste have five volume (not more...)

Here is the dataset :

                                                                             TRKS   %Used  xt
         XXXXXX.YYYY.ZZZZZZ.AAA                             22500  100     4 
         XXXXXX.YYYY.ZZZZZZ.AAA                             45000  100     2 
         XXXXXX.YYYY.ZZZZZZ.AAA                             45000  100     1 
         XXXXXX.YYYY.ZZZZZZ.AAA                             45000  100     2 
         XXXXXX.YYYY.ZZZZZZ.AAA                            45000  100     2 
***************************** End of Data Set list ****************************

and the allocation is (3000,1500).

The error message is :


Thanks for youy help.
Posts: 408
Joined: Mon Nov 16, 2009 7:24 pm
Has thanked: 1 time
Been thanked: 0 time

Re: how to allocate very big dataset

Postby Akatsukami » Fri Dec 16, 2011 8:32 pm

And yet the message referred to an EOV abend on a tape device. Would you care to explain?
"You have sat too long for any good you have been doing lately ... Depart, I say; and let us have done with you. In the name of God, go!" -- what I say to a junior programmer at least once a day
User avatar
Global moderator
Posts: 1058
Joined: Sat Oct 16, 2010 2:31 am
Location: Bloomington, IL
Has thanked: 6 times
Been thanked: 51 times

Re: how to allocate very big dataset

Postby Robert Sample » Fri Dec 16, 2011 8:40 pm

Consult your site suport group handling storage management. It may not be possible for your data set to be allocated on 5 volumes. Furthermore, unless you are using a recent version of z/OS and a storage class to allow for large data sets on volumes, specifying (3000,1500) cylinders in your SPACE parameter is useless. If the volumes in question are 3390 mod 3 then 3335 cylinders is their limit, so the secondary could NEVER be allocated. And if the volumes are 3390 mod 9 (or higher), then (3000,1500) would only work when large data set allocation is supported since otherwise the absolute limit for a sequential data set on a disk volume is 65535 tracks or 4359 cylinders.

But the bottom line in all cases like this is to work with your site storage group since they know MUCH more about your environment then we possibly could.
Robert Sample
Global moderator
Posts: 3657
Joined: Sat Dec 19, 2009 8:32 pm
Location: Dubuque, Iowa, USA
Has thanked: 1 time
Been thanked: 270 times

Re: how to allocate very big dataset

Postby steve-myers » Fri Dec 16, 2011 11:49 pm

Mr. Sample may have missed two important details:
  • Dataset size on one volume.
    As Mr. Sample says, a 3330-3 has just 3335 cylinders. Allocating 3000 cylinders effectively fills the volume; the system cannot allocate 1500 more cylinders on the volume
  • The gory details of allocating a dataset on multiple volumes.
    The same z/OS function, "end of volume," is a catch all routine that handles many different errors. The IEC028I message from an S837-xx ABEND covers a number of codes. Most of the codes are for tape, but 08 is for both disk and tape. Unless you do something in your JCL, you can extend a dataset to 5 volumes. Period. End of story. It doesn't matter whether it's disk or tape. If you expect the dataset will be on more than 5 volumes, you must provide a volume limit in JCL in the VOLUME parameter or specify the volume serials in the VOLUME parameter, or both, as discussed in this link, which evidently you did not review. Of course the volumes must be available. If they are not available, you're done for.
    When you extend the dataset to a new volume, the initial allocation is the secondary allocation, 1500 cylinders in your case, and you can get secondary allocation in the usual way.
This dataset probably should be on real, not virtual, tape, but that's just my horseback opinion. As Mr. Sample says, you really should discuss this with your storage management people, not here.
Global moderator
Posts: 2087
Joined: Thu Jun 03, 2010 6:21 pm
Has thanked: 4 times
Been thanked: 235 times

Re: how to allocate very big dataset

Postby samb01 » Mon Dec 19, 2011 3:38 pm

I successed with this allocation :


Posts: 408
Joined: Mon Nov 16, 2009 7:24 pm
Has thanked: 1 time
Been thanked: 0 time

Re: how to allocate very big dataset

Postby dick scherrer » Tue Dec 20, 2011 1:45 am


And exactly how much space was needed to hold the data? Sounds like the original "Hugeness" of the file was questionable?

Good to hear it is working (and thanks for letting us know :) ), but i'd make sure i understand exactly Why this worked and that there is no surprise at a later date.
Hope this helps,
User avatar
dick scherrer
Global moderator
Posts: 6269
Joined: Sat Jun 09, 2007 8:58 am
Has thanked: 3 times
Been thanked: 93 times

Re: how to allocate very big dataset

Postby deemis » Tue Mar 06, 2012 8:10 pm

hi guys,
i need help in space parameter for jcl.
is there any tool or formula available to calculate? for e.g record is 16169902 lrec=339 . so what is space is being used for this. and what are the prmary and secondary space.
Posts: 2
Joined: Fri Feb 17, 2012 4:07 pm
Has thanked: 0 time
Been thanked: 0 time

Re: how to allocate very big dataset

Postby prino » Tue Mar 06, 2012 8:19 pm

16169902 / ((27998 div 339) * 339) + 1 tracks

And don't fecking tag on to an existing thread, create a new one!
Robert AH Prins
robert.ah.prins @ the.17+Gb.Google thingy
User avatar
Posts: 613
Joined: Wed Mar 11, 2009 12:22 am
Location: Vilnius, Lithuania
Has thanked: 3 times
Been thanked: 27 times

Return to JCL


  • Related topics
    Last post