I know this message is old, but I've read this a number of times and wanted
to comment on it.
I remember when I used to work more with "another" backup product, they had
an NT registry hack that worked with adaptec SCSI cards that could increase
the max block size to 128k.
Has anyone heard of this?
Is it possible that NBU has hard-coded the block size to 64k, so that this
registry hack would not help at all?
At 01:08 PM 12/28/00 -0800, Bob Bakh wrote:
>One thing you all might want to keep in mind is that your max on NT is a 64k
>buffer. I usually set it to 128 on a UNIX server, but if it is also running
>Oracle or any other RDBMS, you may be in trouble with semaphores and shared
>memory segments. you may have to play with the /etc/system file or whatever
>it is in your case to get it to work, you will normally get an status 11
>error. This is with most DLT 7000 drives.
>
>On faster drives you can play with buffers of 128k 256k or 512, we don't see
>much performance improvement over 256k
>
>Good luck.
>
>Bob
>
>-----Original Message-----
>From: David A. Chapa [mailto:david AT datastaff DOT com]
>Sent: Monday, December 18, 2000 10:30 AM
>To: Jim VandeVegt
>Cc: NetBackup Veritas
>Subject: Re: [Veritas-bu] NUMBER_ & SIZE_DATA BUFFERS
>
>
>Jim:
>
>Before I would change the NUMBER and SIZE, I would
>create a bptm logging directory.
>
>/usr/openv/netbackup/logs/bptm
>
>this should be created on each of your media servers
>(including any master that has netbackup controlled
>devices).
>
>this log will let you know how your buffer situation
>looks.
>
>For one of my clients, I have found that the defaults
>were just about fine, with a slightly different
>orientation.
>
>instead of the 8x64 I tried 16x32 (most of the backups
>use multiplexing). This increased the speed quite a
>bit and using the bptm logging directory you are able
>to measure those results accordingly.
>
>My 2 cents.
>
>David Chapa
>
>Quoting Jim VandeVegt <vandevegt AT yahoo DOT com>:
>
> > Can other readers share what your NUMBER and
>SIZE_DATA_BUFFERS are in
> > /usr/openv/netbackup/db/config?
> >
> > My master server is a Sun 450, 1 GB RAM, 2x 400 MHz
>processing.
> > Shared Memory MAX is set to 1GB in /etc/system. Six
>tape drives.
> >
> > I recently noticed my backups were going pretty slow
>and I realized I
> > didn't have these two files in place, thus I was
>using the default
> > 8x32KB buffers. I remembered from my NetBackup class
>that almost
> > everyone increases the size of the buffers and
>usually decreases the
> > numbers, but maintains a great increase in the total
>size of the
> > buffer (size times number).
> >
> > I tried 4 buffers 1048576 bytes (2^20 = 1 MB) big. I
>noticed a
> > speedup in backups, but some NT clients started
>throwing errors.
> >
> > Most of the errors are 174, a media manager error,
>which doesn't make
> > a lot of sense; and a few error 41 (network
>connection timeout).
> >
> > But I haven't fully diagnosed the errors net in the
>log files. I'm
> > more interested in what other installations use for
>the buffer settings.
> >
> > =====
> > Jim VandeVegt
> > vandevegt AT yahoo DOT com vandevegt AT avaya DOT com
> > http://members.xoom.com/vandevegt
> > Toilet paper should come off the front!
> >
> > __________________________________________________
> > Do You Yahoo!?
> > Yahoo! Shopping - Thousands of Stores. Millions of
>Products.
> > http://shopping.yahoo.com/
> > _______________________________________________
> > Veritas-bu maillist - Veritas-
>bu AT mailman.eng.auburn DOT edu
> >
>http://mailman.eng.auburn.edu/mailman/listinfo/veritas-
>bu
> >
>
>
>
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>David A. Chapa
>Consulting Manager
>DataStaff, Inc.
>847 413 1144
>_______________________________________________
>Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
>http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
>_______________________________________________
>Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
>http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
|
