Joined: 06 Jun 2008 Posts: 8505 Location: Dubuque, Iowa, USA
DATACLAS requires SMS to be running while VOLUME= is a DCB parameter that applies whether or not SMS is running. DATACLAS can be used to specify a number of data set values besides just MULTIVOL.
Quote:
Which one to go for ?
If your site has a standard, follow it. If your site has SMS running, and the other parameters of the DATACLAS are compatible with your desired data set, use the DATACLAS -- otherwise, use VOLUME=.
Joined: 30 Nov 2013 Posts: 772 Location: The Universe
Robert Sample wrote:
... VOLUME= is a DCB parameter that applies whether or not SMS is running. DATACLAS can be used to specify a number of data set values besides just MULTIVOL.
Quote:
Which one to go for ?
If your site has a standard, follow it. If your site has SMS running, and the other parameters of the DATACLAS are compatible with your desired data set, use the DATACLAS -- otherwise, use VOLUME=.
No. VOLUME=(,,,n) is a JCL parameter. It has nothing to do with the DCB. It defines the maximum number of volumes that can be specified - usually created - for the allocation. It does not have to be specified for an existing data set as it is implied by the number of volumes explicitly specified for the data set or specified in the catalog for the data set. The default is 5.
I know little about how DATACLAS and volume limits interact. Perhaps someone with a more detailed knowledge of SMS can help us here. Without more detailed knowledge about this, if I was concerned the data set would require more than 5 volumes I'd specify VOLUME=(,,,n) in my JCL .
Best way to go for Multivolume is by DATACLAS as this allows the use of several Space Constraint Relief methods to avoid an X37 ABEND. In addition the jobs will not fill up the catalog with Candidate Volumes as this would be the case if a static VOL=(,,,<n>) parm is coded in JCL. However this requires DVC to be enabled in the DATACLAS.
There is only one thing to consider if you have lots of datasets and a high DVC value set, that's the TIOT. But error messages will tell you if that's the case.
Joined: 30 Nov 2013 Posts: 772 Location: The Universe
Joerg.Findeisen wrote:
... the jobs will not fill up the catalog with Candidate Volumes as this would be the case if a static VOL=(,,,<n>) parm is coded in JCL. However this requires DVC to be enabled in the DATACLAS.
There is only one thing to consider if you have lots of datasets and a high DVC value set, that's the TIOT. But error messages will tell you if that's the case.
I thought candidate volumes were for VSAM. Is space reserved for candidate volumes for non VSAM??? I don't know!!! In any event, the space usage for a candidate volume is trivial, even with thousands of data sets having space reserved for candidate volumes is unlikely to blow out the catalog space.
TIOT space is required for devices, not volumes. Of course with disk, each device / volume requires TIOT space. Even with multiple volume tapes, it's probably silly to provide one drive for each volume. Two drives is usually enough for tape, and even that's over doing it with robots mounting the tapes!
1) You can add candidates to NVSAM datasets with IDCAMS AVOL(* ..), this works although it's static and will definitely stress your catalog. The catalog is stressed by this with (longer records) inserts and deletes and this might be one of the reasons why the option CA Reclaim was introduced. Conclusion, Do not use - consider Dynamic Volume Count (aka DVC). Makes more sense, less work for the Storage Monkey.
2) TIOT has devices, right. With 64K you get (TIOT SIZE: 64 (MAX DDS: 3273)) leading to a maximum of nearly 20 volumes per DD if used up to the full. Hence I said, do not over use DVC and watch messages. In my shop some application hit the limit.
Joined: 30 Nov 2013 Posts: 772 Location: The Universe
Joerg.Findeisen wrote:
...TIOT has devices, right. With 64K you get (TIOT SIZE: 64 (MAX DDS: 3273)) leading to a maximum of nearly 20 volumes per DD if used up to the full. Hence I said, do not over use DVC and watch messages. In my shop some application hit the limit.
Yes and no. It has 24 bytes of fixed data followed by variable data. It has one entry for every DD statement. These entries vary in size; the principle determination of their size is the number of devices allocated to the DD statement.
It is really improper to say a TIOT can have xx DD statements because of the variable size of each DD entry. When you say xx DD statements you are making the assumption each DD statement has one device. Not safe.