In <ADSM-L%1998032517240056 AT VM.MARIST DOT EDU>, on 03/26/98
   at 09:28 AM, "Nicholas, Murray, Haltek/AU" <MurrayNi AT HALTEK.COM DOT AU> 
said:

>Allen,

>When I read your solution I thought it was brilliant in its simplicity.
>Then I read the "ADSM for MVS Admin Guide" under the heading "Assigning
>Random Access Storage Pools".  The full text of this section of the manual
>says
>        "Volumes in random access storage pools must be predefined."

>Unfortunately, unless you can demonstrate an error in the documentation or
>a recent change, it would appear that you cannot allocate scratch volumes
>to disk-based storage pools.

>Given the specific problem and the conceptual validity of the solution, it
>would seem to me that there is a valid basis for this becoming a
>requirement for IBM to modify the code.

>Murray Nicholas
>IT Systems Consultant
>Galeforce Information Services Pty Ltd
>email:  murray AT galeforce.bu.aust DOT com

I think there is simplicity in the current design and it shouldn't be
changed. Having all the diskpool volumes predefined allows the server to
quickly find a place to put the files coming up the wire.  This is where
response time needs to be quickest to get the nightly backup load done.

If a requirement is going to be submitted, I have a suggestion which is
simpler still.  Allow the node definition to have a nextpool field which
overrides the default nextpool in the storagepool definition when migration
and reclaim are running.

--
-----------------------------------------------------------
-----------------------------------------------------------
Bill Colwell
Bill Colwell
C. S. Draper Lab
Cambridge, Ma
bcolwell AT draper DOT com
-----------------------------------------------------------
=========================================================================