[Veritas-bu] Multiplexing

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01C200E0.95093C80
Content-Type: text/plain

Hi all,
I had a serious problem in my duplications for offsite storage and the
ever-increasing time to copy the 12 images or so I send offsite weekly.
(search the archive for the threads if you want).

You could see the negative effects of multiplexing on my iostat checks.  For
every 10K I'd read, I'd write 10%-50% of that.  I was spinning a lot of tape
to read a little data.

Anyway - my solution was to break my backups into two classes, a multiplexed
one and a demux'd one.  The demux'd backups are those systems that need to
be duplicated.  The time to duplicate under the old method varied a lot by
how the race to grab drives ended up interleaving the images but in general
I moved from 3-5 days to duplicate to just about 1 day for the same set of
data.

To answer the question - I use 4096M as a tape fragment size.

$.02
-Mark

-----Original Message-----
From: Richard Hellier [mailto:rlh AT lsil DOT com]
Sent: Tuesday, May 21, 2002 9:26 AM
To: Alvarez, Martin BGI SDC
Cc: 'veritas-bu AT mailman.eng.auburn DOT edu'
Subject: Re: [Veritas-bu] Multiplexing


Martin,
        When you saw the l-o-n-g restore times, did you also have
a non-zero fragment size set?


Cheers,

Richard.

"Alvarez, Martin BGI SDC" wrote:
> 
> David..  I agree with your comments.  We have a serious reliance on
> restoring data from tape rather then grow our online copies to a
sufficient
> level where we can do disk restores.  That being said, when we tested
> multiplexing we saw almost a doubling of the restore time from what it
> normally took in a single stream backup.
> This was unacceptable to the user community so MPX was not an option for
us.
> 
> Martin

------_=_NextPart_001_01C200E0.95093C80
Content-Type: text/html
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3DUS-ASCII">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2653.12">
<TITLE>RE: [Veritas-bu] Multiplexing</TITLE>
</HEAD>
<BODY>

<P><FONT SIZE=3D2>Hi all,</FONT>
<BR><FONT SIZE=3D2>I had a serious problem in my duplications for =
offsite storage and the ever-increasing time to copy the 12 images or =
so I send offsite weekly. (search the archive for the threads if you =
want).</FONT></P>

<P><FONT SIZE=3D2>You could see the negative effects of multiplexing on =
my iostat checks.&nbsp; For every 10K I'd read, I'd write 10%-50% of =
that.&nbsp; I was spinning a lot of tape to read a little =
data.</FONT></P>

<P><FONT SIZE=3D2>Anyway - my solution was to break my backups into two =
classes, a multiplexed one and a demux'd one.&nbsp; The demux'd backups =
are those systems that need to be duplicated.&nbsp; The time to =
duplicate under the old method varied a lot by how the race to grab =
drives ended up interleaving the images but in general I moved from 3-5 =
days to duplicate to just about 1 day for the same set of =
data.</FONT></P>

<P><FONT SIZE=3D2>To answer the question - I use 4096M as a tape =
fragment size.</FONT>
</P>

<P><FONT SIZE=3D2>$.02</FONT>
<BR><FONT SIZE=3D2>-Mark</FONT>
</P>

<P><FONT SIZE=3D2>-----Original Message-----</FONT>
<BR><FONT SIZE=3D2>From: Richard Hellier [<A =
HREF=3D"mailto:rlh AT lsil DOT com">mailto:rlh AT lsil DOT com</A>]</FONT>
<BR><FONT SIZE=3D2>Sent: Tuesday, May 21, 2002 9:26 AM</FONT>
<BR><FONT SIZE=3D2>To: Alvarez, Martin BGI SDC</FONT>
<BR><FONT SIZE=3D2>Cc: 'veritas-bu AT mailman.eng.auburn DOT edu'</FONT>
<BR><FONT SIZE=3D2>Subject: Re: [Veritas-bu] Multiplexing</FONT>
</P>
<BR>

<P><FONT SIZE=3D2>Martin,</FONT>
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>When you =
saw the l-o-n-g restore times, did you also have</FONT>
<BR><FONT SIZE=3D2>a non-zero fragment size set?</FONT>
</P>
<BR>

<P><FONT SIZE=3D2>Cheers,</FONT>
</P>

<P><FONT SIZE=3D2>Richard.</FONT>
</P>

<P><FONT SIZE=3D2>&quot;Alvarez, Martin BGI SDC&quot; wrote:</FONT>
<BR><FONT SIZE=3D2>&gt; </FONT>
<BR><FONT SIZE=3D2>&gt; David..&nbsp; I agree with your comments.&nbsp; =
We have a serious reliance on</FONT>
<BR><FONT SIZE=3D2>&gt; restoring data from tape rather then grow our =
online copies to a sufficient</FONT>
<BR><FONT SIZE=3D2>&gt; level where we can do disk restores.&nbsp; That =
being said, when we tested</FONT>
<BR><FONT SIZE=3D2>&gt; multiplexing we saw almost a doubling of the =
restore time from what it</FONT>
<BR><FONT SIZE=3D2>&gt; normally took in a single stream backup.</FONT>
<BR><FONT SIZE=3D2>&gt; This was unacceptable to the user community so =
MPX was not an option for us.</FONT>
<BR><FONT SIZE=3D2>&gt; </FONT>
<BR><FONT SIZE=3D2>&gt; Martin</FONT>
</P>

</BODY>
</HTML>
------_=_NextPart_001_01C200E0.95093C80--