Re: Export quesion
2004-08-26 20:12:02
Mark,
<one's_opinion_mode>
Any machine that has 3.7TB worth of files is *begging* to be broken up
into multiple node names, each node responsible for backing up a portion
of a physical machine. Particularly if you're using slow tape technology
like DLT.
</one's_opinion_mode>
I agree with you, and in fact, we've done that with our main file servers.
But this is a Domino TDP node - all the databases (about 1000) and the
entire history of its transaction log.
I suppose I could have exported the databases and logs separately, but
that would have just turned a nine-day export into two five-day exports.
It should be possible for TSM to do a checkpoint at the end of every
filled volume.
As for our DLTs, they are slower than LTOs, but when we bought the 4/40
three years ago, it was a "Cadillac" compared to our old one-drive DLT
4000 autoloader. Our offsite tape throughput went from 2.5 MB/s to 32
MB/s. And three years ago, LTO-1 was just being introduced. We didn't
think they were a suitable DR medium because of that newness. We only use
DLT for offsite. Onsite we use LTO-1 and 3570, and hope to begin using a
VTL by next year.
Thanks for your feedback.
Tab Trepagnier
TSM Administrator
Laitram, L.L.C.
"Stapleton, Mark" <mark.stapleton AT BERBEE DOT COM>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
08/26/2004 06:55 PM
Please respond to "ADSM: Dist Stor Manager"
To: ADSM-L AT VM.MARIST DOT EDU
cc:
Subject: Re: Export quesion
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On
Behalf Of Tab Trepagnier
>This is an opportunity for an enhancement. I've been bit by
>that situation also.
>
>The problem is that an export is a single transaction. Either
>it fails or it succeeds. As I write this I'm holding my
>breath hoping that the last 100 GB of the current _3.7 TB_
>export complete without a problem. It has been running for
>eight days now and has survived two "dropped tape"
>events by our 3583 LTO and is currently writing DLT # 64 in
>our HP 4/40.
<one's_opinion_mode>
Any machine that has 3.7TB worth of files is *begging* to be broken up
into multiple node names, each node responsible for backing up a portion
of a physical machine. Particularly if you're using slow tape technology
like DLT.
</one's_opinion_mode>
Exports have been notoriously slow to write (and read) from ADSM/TSM
first iteration. (This is one reason why you can now export directly
from one TSM server to another.) Did you perform a MOVE NODEDATA prior
to the export? If you didn't, and the node is not collocated, you made
your life far harder than it had to be.
--
Mark Stapleton (stapleton AT berbee DOT com)
Berbee Information Networks
Office 262.521.5627
|
|
|