"Butt, Jawad (CAP, ITS, CA)" wrote:

> Yes well its a pity because if something can go down, it will go down.
> So you can put mirrored OS in a system, redundant power, fans, service
> processors etc etc..but if the main processor went down or memory or the
> backplane etc etc..you cant do anything ..
>
> now in case of ADSM , with the backup windows being so small and all, if
> that happens just before or around that time, you are basically without
> one days backup until the next night..
> HACMP will transfer your volume groups and ip addresses but in casse of
> ADSM u need more than that..u need the ADSM server running with all the
> customised files and devices...
> I wish IBM comes up witha  preferred solution to this
>

I've designed and configured an ADSM server HACMP solution on an SP with
3590/3494.  We were even able to have the server fail over in the middle of
a a client backup and have that backup continue once the server is up on the
other node.  We had to make sure that we configured the server nodes so that
failover was possible and write sensible start and stop scripts, just like
you would with any other application.

That is just a fact of IBM's HACMP; it does not come with pre-canned
Application Servers (other than specific support for NFS, viewing NFS as a
service).  As a corollary, ADSM server runs on many platforms, some of which
have different HA software solutions (HP/UX with ServiceGuard, NT with
Wolfpack).  It would be a non-trivial to come up with a standard scheme that
would work in all those environments.

That is what consultants are for :-) or your own smart people.

Cheers...
Bruce
--
Bruce Elrick, Ph.D.
Bruce Elrick, Ph.D.
belrick AT home DOT com
http://members.home.net/belrick/
ADSM, AIX Support, SP, and HACMP Certified.