ADSM-L

Re: CAN RESTORE ACTIVITY BE TRACED?

2006-07-27 15:26:31
Subject: Re: CAN RESTORE ACTIVITY BE TRACED?
From: "Allen S. Rout" <asr AT UFL DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Thu, 27 Jul 2006 15:23:02 -0400
>> On Thu, 27 Jul 2006 10:37:46 -0500, Laura Mastandrea <lmastand AT 
>> CHOOSEBROADSPIRE DOT COM> said:


> Running TSM 5.2.4 on AIX 5.3 and restoring a Windows 2000 Advanced
> server client.  The filespace is on tape in a non-colocated storage
> pool.  The tape library is a 3584 with LTO3 drives.  It takes about
> 3 hours to restore 7 GB of compressed data.

> I can monitor by doing q se f=d, but for three hours I can't do
> this.  The server at the time was not busy with other tape activity
> that I could see.

> Is there a trace I can run on this a restore to see what is taking
> so long?  Is there a SQL Select that would give me where the time
> was spent for this restore or is my only source the activity log?


What did the network throughput at the end of the restore look like?
Are you certain that everything was properly configured?  The single
most common problem I've encountered with throughput is the "New box,
duplex mismatch, slow network" problem.

On the AIX side, you can use nmon to watch what the network is doing,
and from that you could see wether the flow was steady or spiky.  I
have no idea how one monitors a windows box.

Unless you've got some serious disk under your windows box, the LTO
streaming speed is going to blow the doors off everything in the
critical path between tape and client disk.  This will mean that
you'll do a lot of backhitching, which I understand LTO is not so good
at.  If you see spiky network on the server side, then that's a decent
bet.


- Allen S. Rout

<Prev in Thread] Current Thread [Next in Thread>