Re: Suggestion for ADSM enhancement to reduce the need for collocation
1998-08-28 09:50:00
Subject: |
Re: Suggestion for ADSM enhancement to reduce the need for collocation |
From: |
Francis Maes <fr.maes AT CGER DOT BE> |
Date: |
Fri, 28 Aug 1998 15:50:00 +0200 |
Hello Mike,
I am 100% agree with you about ADSM restore enhancements.
With or without collocation, a "big" node will allways have his data spread
on many cartridges (3490).
It would also be a very good thing to have collocation in DFSMShsm (think to
ABARS) but this is another story.
Best regards,
Francis.
-----Message d'origine-----
De : Mike Wilson <mwilson6 AT TUELECTRIC DOT COM>
De : Mike Wilson <mwilson6 AT TUELECTRIC DOT COM>
À : ADSM-L AT VM.MARIST DOT EDU <ADSM-L AT VM.MARIST DOT EDU>
Date : vendredi 21 août 1998 16:56
Objet : Suggestion for ADSM enhancement to reduce the need for collocation
>I have seen discussions on this list in the past talking about the need to
>reduce the number of tape mounts on restore. They usually start out by
>asking can ADSM perform full backups? Which in turn usually leads to
>recommendations of collocating tapes. Then it is discussed how collocation
>can lead to greatly increased tape mounts, tape wear & less efficient tape
>utilization and is up to the admin. to weigh the benefits versus the cost.
>What I am wondering is has it ever been suggested to lessen the impact of
>data being spread over multiple tapes by making a couple of enhancements to
>ADSM.
>
>1. Since the bottleneck in many environments is usually the network, why
>not have an option to use the disk storage pool space as a buffer for the
>restore? ADSM would load the first tape and restore directly to the disk on
>the server at disk & tape speeds. It could then go on to the next tape
>loading data to the disk storage pool without waiting while data from the
>1st tape is still being transferred to the client.
>
>2. Multitask the restore process. DFHSM on OS/390 can restore an entire
>DASD volume from an all incremental backup. Data could be spread out over
>many tapes and could require a great many mounts to restore the volume.
>This is very similar to ADSMs incremental without collocation. But hsm does
>not perform collocation of this data, instead it was enhanced to allow the
>admin. to specify how many concurrent restore task could run (up to 64,
>provided you have that many tape drives).
>
>It seems to me that these enhancements could allow the ADSM node to see a
>constant stream of data coming across the wire during restore. The client
>would not see all this recovery process happening in the back ground. Some
>ADSM sites may then have an option to turn off collocation. Have these
>improvements already been suggested? I would like to hear from others what
>there opinion is on these.
>
>Thanks,
>Mike Wilson
>
|
|
|