While best practice is to let the system choose, if you insist on using 
VOL=SER=foo for a temporary, it will work. IMHO, it's best to let SMS do its 
thing.

--
Shmuel (Seymour J.) Metz
http://mason.gmu.edu/~smetz3
עַם יִשְׂרָאֵל חַי



________________________________________
From: IBM Mainframe Discussion List <[email protected]> on behalf of Jon 
Perryman <[email protected]>
Sent: Thursday, November 23, 2023 1:53 PM
To: [email protected]
Subject: Re: UNIX REXX LINKMVS TASKLIB?

On Thu, 23 Nov 2023 01:47:18 -0600, Paul Gilmartin <[email protected]> wrote:
>>On Thu, 23 Nov 2023 04:18:05 +0000, Farley, Peter wrote:

>>In this day and age, why in the world would you evec contemplate using a 
>>non-cataloged library?
>>
>In parallel with the JCL Ref. mentioning use of temporary data sets as link 
>libraries:
><https://www.ibm.com/docs/en/zos/3.1.0?topic=statement-pgm-parameter>
>    Use the PGM parameter ..., or a temporary library.

Temporary datasets never use VOL=SER=. They are accessed through a DD that 
optionally exists for the life of the job (DDNAME= or DSN=&&amp;XXX). 
Technically they are uncatalogued but the word uncatalogued implies use of 
VOL=SER=,

>>Performance was admirable and acceptable but this usage was obviously not a 
>>high-volume commercial application, so YMMV.
>>
>I was just taking to heart the advice in
><https://www.ibm.com/docs/en/zos/3.1.0?topic=shell-controlling-use-steplibs>
>    ... To improve performance of the z/OS® shell, avoid propagating STEPLIBs 
> ...

Think of this in terms of z/OS. USS STEPLIB variable is actually an attach 
tasklib (not a true step). The problem occurs when you FORK() a process which 
must allocate a new unique DDN, open it and read the directories during ATTACH. 
Much of UNIX is built around FORKing is cheap. For instance consider MAKE 
(compile) which could potentially do a 100,000 FORKs in a short amount of time.

Performance is in the eye of the beholder. Performance is more than one thing 
that runs extremely fast. z/OS can run at 100% busy without impacting users. If 
I run MAKE to compile in LINUX, it is treated the same as if I'm using edit.  
Performance must be measured in overall impact which can vary greatly by 
environment.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to