[Veritas-bu] tape drive performance on image copies

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_01C24EA9.16BC8B90
Content-Type: text/plain

Check your buffer settings - this made a difference for me.  See this tech
note:
http://seer.support.veritas.com/docs/183702.htm.

Having my NET_BUFFER_SZ set the same as my SIZE_DATA_BUFFERS made a
measurable difference in my duplication speed.

Also, consider the source images.  If they're muliplexed then the desired
image must be sorted from the undesired blocks resulting in slower copies.

HTH - Mark

-----Original Message-----
From: Eric Forgette [mailto:epforgette AT overnite DOT com]
Sent: Wednesday, August 28, 2002 9:15 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] tape drive performance on image copies


I wrote a perl script to duplicate the most recent, successful, full
backup images for a particular class.  The script is available at
http://perlvault.sourceforge.net/ (beta release still...)

The problem I am having is in the performance of the image copies.  My
expectation was that the tape to tape copies would be faster than the
backups themselves.  The regular backup performance is limited by the
fact that there are tons of small files and by the network speed / cpu
speed of the client.  My understanding is that the images (once on tape)
are one sequential file, which should lend itself well to copying.

The master is an E4500 running Solaris 5.8.  I am using LTO drives which
are SAN attached.  I've seen throughput (using iostat) of 11 MB/second
during regular backups.  During the image copies, I get between 5 - 7
MB/second.  The regular backups do not use any multiplexing.

Is there any tuning parameters I can tweak?  
Anyone have any ideas?

Regards,
-Eric
-- 
Eric Forgette
Unix Systems Administrator
_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu

------_=_NextPart_001_01C24EA9.16BC8B90
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] tape drive performance on image copies</TITLE>
</HEAD>
<BODY>

<P><FONT SIZE=3D2>Check your buffer settings - this made a difference =
for me.&nbsp; See this tech note:</FONT>
<BR><FONT SIZE=3D2><A =
HREF=3D"http://seer.support.veritas.com/docs/183702.htm"; =
TARGET=3D"_blank">http://seer.support.veritas.com/docs/183702.htm</A>.</=
FONT>
</P>

<P><FONT SIZE=3D2>Having my NET_BUFFER_SZ set the same as my =
SIZE_DATA_BUFFERS made a measurable difference in my duplication =
speed.</FONT>
</P>

<P><FONT SIZE=3D2>Also, consider the source images.&nbsp; If they're =
muliplexed then the desired image must be sorted from the undesired =
blocks resulting in slower copies.</FONT></P>

<P><FONT SIZE=3D2>HTH - Mark</FONT>
</P>

<P><FONT SIZE=3D2>-----Original Message-----</FONT>
<BR><FONT SIZE=3D2>From: Eric Forgette [<A =
HREF=3D"mailto:epforgette AT overnite DOT com">mailto:epforgette AT overnite 
DOT com</=
A>]</FONT>
<BR><FONT SIZE=3D2>Sent: Wednesday, August 28, 2002 9:15 AM</FONT>
<BR><FONT SIZE=3D2>To: veritas-bu AT mailman.eng.auburn DOT edu</FONT>
<BR><FONT SIZE=3D2>Subject: [Veritas-bu] tape drive performance on =
image copies</FONT>
</P>
<BR>

<P><FONT SIZE=3D2>I wrote a perl script to duplicate the most recent, =
successful, full</FONT>
<BR><FONT SIZE=3D2>backup images for a particular class.&nbsp; The =
script is available at</FONT>
<BR><FONT SIZE=3D2><A HREF=3D"http://perlvault.sourceforge.net/"; =
TARGET=3D"_blank">http://perlvault.sourceforge.net/</A> (beta release =
still...)</FONT>
</P>

<P><FONT SIZE=3D2>The problem I am having is in the performance of the =
image copies.&nbsp; My</FONT>
<BR><FONT SIZE=3D2>expectation was that the tape to tape copies would =
be faster than the</FONT>
<BR><FONT SIZE=3D2>backups themselves.&nbsp; The regular backup =
performance is limited by the</FONT>
<BR><FONT SIZE=3D2>fact that there are tons of small files and by the =
network speed / cpu</FONT>
<BR><FONT SIZE=3D2>speed of the client.&nbsp; My understanding is that =
the images (once on tape)</FONT>
<BR><FONT SIZE=3D2>are one sequential file, which should lend itself =
well to copying.</FONT>
</P>

<P><FONT SIZE=3D2>The master is an E4500 running Solaris 5.8.&nbsp; I =
am using LTO drives which</FONT>
<BR><FONT SIZE=3D2>are SAN attached.&nbsp; I've seen throughput (using =
iostat) of 11 MB/second</FONT>
<BR><FONT SIZE=3D2>during regular backups.&nbsp; During the image =
copies, I get between 5 - 7</FONT>
<BR><FONT SIZE=3D2>MB/second.&nbsp; The regular backups do not use any =
multiplexing.</FONT>
</P>

<P><FONT SIZE=3D2>Is there any tuning parameters I can tweak?&nbsp; =
</FONT>
<BR><FONT SIZE=3D2>Anyone have any ideas?</FONT>
</P>

<P><FONT SIZE=3D2>Regards,</FONT>
<BR><FONT SIZE=3D2>-Eric</FONT>
<BR><FONT SIZE=3D2>-- </FONT>
<BR><FONT SIZE=3D2>Eric Forgette</FONT>
<BR><FONT SIZE=3D2>Unix Systems Administrator</FONT>
<BR><FONT =
SIZE=3D2>_______________________________________________</FONT>
<BR><FONT SIZE=3D2>Veritas-bu maillist&nbsp; -&nbsp; =
Veritas-bu AT mailman.eng.auburn DOT edu</FONT>
<BR><FONT SIZE=3D2><A =
HREF=3D"http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu"; =
TARGET=3D"_blank">http://mailman.eng.auburn.edu/mailman/listinfo/veritas=
-bu</A></FONT>
</P>

</BODY>
</HTML>
------_=_NextPart_001_01C24EA9.16BC8B90--