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_01C27AB0.2E897BB0
Content-Type: text/plain;
charset="iso-8859-1"
Increasing the buffer settings to 256K for our LTO drives resulted in a
nearly doubling of backup speed so, yes, it can have a measurable effect.
Matching SIZE_DATA_BUFFERS to NET_BUFFER_SZ was also beneficial.
Also set the NET_BUFFER_SZ to match on the clients.
I haven't played with NUMBER_DATA_BUFFERS_RESTORE.
-Mark
-----Original Message-----
From: Gravizi, Thomas [mailto:thomas_gravizi AT steris DOT com]
Sent: Wednesday, October 23, 2002 10:01 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] Improving Restore Times?
Greetings,
We have issues performing restores of many small files. We are performing
cold backups for our Oracle 11i instances which are about 95 Gig and include
about 575k files. The backup times is about 3.5-4 hours, which really isn't
an issue. But, when we try doing a restore, it take between 9-11 hours. I
do have the Veritas NetBackup Performance Tuning white papers v. 1.51, and
have a few questions about that. First of all, our indexing level is
already set to 9, so that setting is confirmed. My next settings would be
the buffer settings. Has any tried changing these and seen results? I
looked at the SIZE_DATA_BUFFERS and for some reason they were set to 262144
or 256k. I changed them (according to the documentation) to 65536 or 64k.
I noticed that the 256k buffer was also on the media server as well. Do the
buffer sizes need to be identical for both master, media, and client? What
about changes to the NUMBER_DATA_BUFFERS and NUMBER_DATA_BUFFERS_RESTORE?
Thanks,
Thomas Gravizi
Systems Administrator
STERIS Corporation - Mentor, OH
thomas_gravizi AT steris DOT com
440.392.7630 - Tel
440.350.7078 - Fax
------_=_NextPart_001_01C27AB0.2E897BB0
Content-Type: text/html;
charset="iso-8859-1"
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<TITLE>Improving Restore Times?</TITLE>
<META content="MSHTML 5.50.4919.2200" name=GENERATOR></HEAD>
<BODY>
<DIV><SPAN class=923091816-23102002><FONT face=Arial color=#0000ff
size=2>Increasing the buffer settings to 256K for our LTO drives resulted in a
nearly doubling o<FONT face=Tahoma color=#000000>f backup speed so, yes, it can
have a measurable effect. Matching SIZE_DATA_BUFFERS to
NET_BUFFER_SZ
was also beneficial. </FONT></FONT></SPAN></DIV>
<DIV><SPAN class=923091816-23102002><FONT face=Tahoma
size=2></FONT></SPAN> </DIV>
<DIV><SPAN class=923091816-23102002><FONT face=Tahoma size=2>Also set the
NET_BUFFER_SZ to match on the clients.</FONT></SPAN></DIV>
<DIV><SPAN class=923091816-23102002><FONT face=Tahoma
size=2></FONT></SPAN> </DIV>
<DIV><SPAN class=923091816-23102002><FONT face=Tahoma size=2>I haven't played
with NUMBER_DATA_BUFFERS_RESTORE.</FONT></SPAN></DIV>
<DIV><SPAN class=923091816-23102002><FONT face=Tahoma
size=2></FONT></SPAN> </DIV>
<DIV><SPAN class=923091816-23102002><FONT face=Tahoma
size=2>-Mark</FONT></SPAN></DIV>
<DIV><SPAN class=923091816-23102002></SPAN><FONT face=Tahoma><FONT size=2><SPAN
class=923091816-23102002><FONT face=Arial
color=#0000ff> </FONT></SPAN></FONT></FONT></DIV>
<DIV><FONT face=Tahoma><FONT size=2><SPAN
class=923091816-23102002> </SPAN>-----Original
Message-----<BR><B>From:</B>
Gravizi, Thomas [mailto:thomas_gravizi AT steris DOT com]<BR><B>Sent:</B>
Wednesday,
October 23, 2002 10:01 AM<BR><B>To:</B>
veritas-bu AT mailman.eng.auburn DOT edu<BR><B>Subject:</B> [Veritas-bu]
Improving
Restore Times?<BR><BR></DIV></FONT></FONT>
<BLOCKQUOTE><!-- Converted from text/rtf format -->
<P><FONT face=Arial size=2>Greetings,</FONT> </P>
<P><FONT face=Arial size=2>We have issues performing restores of many small
files. We are performing cold backups for our Oracle 11i instances
which
are about 95 Gig and include about 575k files. The backup times is
about
3.5-4 hours, which really isn't an issue. But, when we try doing a
restore, it take between 9-11 hours. I do have the Veritas NetBackup
Performance Tuning white papers v. 1.51, and have a few questions about
that. First of all, our indexing level is already set to 9, so that
setting is confirmed. My next settings would be the buffer
settings. Has any tried changing these and seen results? I looked
at the SIZE_DATA_BUFFERS and for some reason they were set to 262144 or
256k. I changed them (according to the documentation) to 65536 or
64k. I noticed that the 256k buffer was also on the media server as
well. Do the buffer sizes need to be identical for both master, media,
and client? What about changes to the NUMBER_DATA_BUFFERS and
NUMBER_DATA_BUFFERS_RESTORE?</FONT></P>
<P><FONT face=Arial size=2>Thanks,</FONT> </P>
<P><B><FONT face=Verdana>Thomas Gravizi</FONT></B> <BR><FONT face=Arial
size=2>Systems Administrator</FONT> <BR><FONT face=Arial size=2>STERIS
Corporation - Mentor, OH</FONT> <BR><FONT face=Arial
size=2>thomas_gravizi AT steris DOT com</FONT> <BR><FONT face=Arial
size=2>440.392.7630 - Tel</FONT> <BR><FONT face=Arial size=2>440.350.7078 -
Fax</FONT> </P><BR></BLOCKQUOTE></BODY></HTML>
------_=_NextPart_001_01C27AB0.2E897BB0--
|