I have been watching this thread and the number of files problem is near
and dear to my heart.
I have one server that has over 2 million files. As far as size it is
only something like 20gb on a 32 gb partition. Routine backup is really
not too much of a problem because the application can manage if a file
is locked by backup - it doesn't use downtime - but it takes almost 24
hours. If there are any other Meditech C/S people out there, you may
recognize this as an EMR server. Nothing more efficient than using an
NT directory structure for a database.
Other than the possibility of catastrophic failure, restores on this are
rare. Writing is done with queued transactions from a process; reading
is through a similar queue passing a file back to a users client.
I am starting the process of upgrading this server from NT4 300mhz 128mb
2 cpu to W2K 3ghz 1gb 2 cpu.
I am looking for options, alternatives, suggestions to back up this
thing, and then restore it on new hardware - and hopefully be
substantially less than the 24 hours + for downtime. It would also not
hurt my feelings to find a longer term resolution for the extended
backup window - I know it can't be good for a tape to be getting data at
floppy disk speeds - or less.
One obvious solution is to do full backup/restore while system is still
on line, then do incremental with the system down. But basically, I'm
chicken. There is very high cost for failure and the way this system is
designed, failure would very likely not be found until too late to abort
the process - I might discover a week later that for whatever reason X
didn't get restored - and there is the possibility that I could do
nothing about it.
Are any of you using Replistore? That seems like it might be a solution
- I will probably load up the evaluation and see what happens but if
others have experience good or bad I would appreciate hearing about it.
-----Original Message-----
From: Narkinsky, Brian [mailto:Brian.Narkinsky AT DEP.STATE.FL DOT US]
Sent: Friday, April 23, 2004 9:26 PM
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Subject: Re: [Networker] Large File system issues
Very large numbers (millions) of files and directories sometimes cause
problems. If you have millions of files I'd look into the Snap Image
modules.
-----Original Message-----
From: Legato NetWorker discussion on behalf of Matt Temple
Sent: Fri 4/23/2004 7:46 PM
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Cc:
Subject: Re: [Networker] Large File system issues
atif wrote:
> hello all.
> one of our business requirements demands a huge filesystem ~
250 GB.
250 gigabytes is now just an off-the-shelf drive that anyone can
buy for
$250.
> Does anyone here have any experience with backups/restores of
a filesystem
> that big?
> Are there any issues anyone can think of that I should
consider?
> Also, we might be able to survive with 80GB fs. Any thoughts
on that?
> Thanks.
>
We backup 60 computers with filesystems that vary between 100 gb
and
1.5 terabytes. No problems with iether the backup or restores.
Paying for the tapes -- priceless!
mht
--
=============================================================
Matthew Temple Tel: 617/632-2597
Director, Research Computing Fax: 617/582-7820
Dana-Farber Cancer Institute mht AT research.dfci.harvard DOT edu
44 Binney Street, LG300/300 http://research.dfci.harvard.edu
Boston, MA 02115 Choice is the Choice!
--
Note: To sign off this list, send a "signoff networker" command
via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
|
