I guess that my SVCDUMP problem sort of triggered this threadlet. The main 
purpose for 'overflow' is when you have a 'proprietary' pool that is more or 
less dedicated to one business unit or function. In our case, SVCDUMPs need to 
be taken to a specific set of volumes so that our RYO dump management process 
can find and process them. Overflow appeared be a simple and convenient way to 
make sure that a verrrry large dump can be taken even though the designated 
pool cannot contain it. Unfortunately, OVERFLOW immediately sucked up all 
SVCDUMPs such that our RYO process never saw any of them. Worst possible 
'solution'. We're now looking at larger SVCDUMP volumes even though 99% of the 
time extra space is unnecessary.

In the application arena, we have some big-gun folks who have historically been 
blessed with their own pools just to make sure that they will always get what 
they need--when they need it. Probably long overdue for a revision in policy, 
but big guns fire big bullets. OVERFLOW seems to resolve the problem pretty 
well because that can service multiple proprietary groups, and those folks 
generally don't care much about volser.   

.
.
J.O.Skip Robinson
Southern California Edison Company
Electric Dragon Team Paddler 
SHARE MVS Program Co-Manager
323-715-0595 Mobile
626-543-6132 Office ⇐=== NEW
[email protected]


-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Cafiero, Tobias M.
Sent: Monday, April 24, 2017 1:51 PM
To: [email protected]
Subject: (External):Re: Who Needs Spill/Overflow Pools anymore?

Team, 
 I agree with the below. There is usually some Testing that doesn't get 
announced or unexpected work load. It's better have extra around, then 
explaining why you have x amount of TB's available, but Applications abended on 
space. 


Tobias Cafiero
Data Resource Management
Core Systems Technology
Lead System Architect
DTCC  New York
Office: 212-855-1117
E-mail: [email protected]
Web: www.dtcc.com <http://www.dtcc.com/

 

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Ronald Kristel
Sent: Monday, April 24, 2017 3:43 PM
To: [email protected]
Subject: Re: Who Needs Spill/Overflow Pools anymore?

What about 'in case of emergencies?'.

(I think) almost all our storagegroups are directed to one substantial big 
overflow storagepool, _just_ for occurances where for whatever reason, 
something starts to heavily use a specific storage pool.

I agree, it's not very efficient to have a massive amount of disks spinning 
without holding any data most of it's lifetime. However, this has classified 
more favorable then facing a time period with space abends.


From my experience, I try to over-provision many of the storagepools with a 
certain amount, so that overflowing does not happen on a regular basis. And I 
monitor/analyse the growth of these storagepools. (And act on adding more 
'MOD54's' when needed).


I think it really depends on what is being stored aswell. Some data is not 
really eligible to be migrated if it's (for example) required for online 
processing. However, data that does not need to meet any criteria in terms of 
'response time' can easily be migrated to save DASD.


I was just wondering, are you using HSM ODM to migrate data?


Ronald Kristel
NL

________________________________
From: IBM Mainframe Discussion List <[email protected]> on behalf of 
Lizette Koehler <[email protected]>
Sent: Monday, April 24, 2017 6:51:38 PM
To: [email protected]
Subject: Who Needs Spill/Overflow Pools anymore?

This is just a discussion topic (Thank you Skip for making me think about this
;-O)

In the past we needed to have them as we were tight on storage.

But today, is that still the case.  What is a good reason to either have or 
don't have SPILL/OVERFLOW pools when we can just add more storage to the pool.

I can add a MOD54 to a pool but that mod54 is not full until I use up all of 
the allotted storage to the MOD54 in the storage array.  So the storage array 
can be over provisioned until I need to go and get management to buy more 
physical storage.

Do I lose anything by having datasets "spill" over to a different pool that may 
not have the same protection as the one it is in?  HSM Backup, Dump, Cleanup 
processes?  Or are there other considerations.

Just asking a question or two.

So basically, why use SPILL/OVERFLOW when you can just add dasd, or do lots of 
migration?

We have the automation tool set up to monitor the pools and if they get too 
full, then start migration on the datasets in that pool.  No manual 
intervention required.



DFHSM and DFSMS do not think like humans when it comes to dataset management.
So there a need to out maneuver them to make datasets go where we want.




Lizette Koehler
statistics: A precise and logical method for stating a half-truth inaccurately


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

Reply via email to