ADSM-L

Storage pool question

1997-04-08 15:59:29
Subject: Storage pool question
From: Eric van Loon <evanloon AT KLM DOT NL>
Date: Tue, 8 Apr 1997 21:59:29 +0200
Hi ADSM-ers!
We currently use a very simple storage pool definition: For every client a 
sequential disk dataset is created and all backup data is written directly to 
this dataset. As soon as the dataset is released by ADSM (because it's full or 
the backup is finished) the SMS ACS routines cause the idle space to be 
released and the dataset is migrated to tape as soon as possible (by standard 
SAMS:DISK jobs).
This works fine, but as soon as a dataset is re-allocated by ADSM it runs into 
a B37 of course. This is not a big problem because the 'volume' obtains the 
read-only status, a new volume is allocated and the operation continues.
My problem is that the method above creates a lot of small files. I am afraid 
that a full restore of one of the filespaces will cause a large amount of 
recalls which will slow down a recovery considerably.
I am thinking about creating some sort of buffer which holds all backup data 
until the amount of data reaches 200 Mb. (for a specific node). As soon as this 
amount is reached ADSM should move all data to another disk dataset which then 
will be migrated to tape by SAMS:DISK later on.
Can somebody please help me with how I should implement this?
Thanks in advance for your reply!
Kindest regards,
Eric van Loon
KLM Royal Dutch Airlines
<Prev in Thread] Current Thread [Next in Thread>
  • Storage pool question, Eric van Loon <=