This is a multi-part message in MIME format.
------_=_NextPart_001_01C268BB.5E564C44
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
I've been trying to tune backups running on a particular media server in =
our environment and am not making much progress. I've modified the =
SIZE_DATA_BUFFERS and NUMBER_DATA_BUFFERS parameters (with local/SAN =
attached tape drives) but it doesn't seem to make much difference. From =
my last test, I saw the following messages in the "bpbkar" log file:
3:13:41.250 PM: [2780.2308] <4> tar_backup::OVPC_EOFSharedMemory: INF - =
bpbkar waited 4198 times for empty buffer, delayed 6071 times
That seems to imply that I don't have enough buffers, but from the =
"bptm" log file I see this:
15:13:41.343 [1976.1984] <2> write_data: waited for full buffer 1608 =
times, delayed 1951 times
which implies that "bpbkar" couldn't fill the buffers fast enough. So =
which is it? In this case, I was doing a very simple, best case =
performance test, backing up a single 4.2GB data file. I'm consistently =
getting about 13MB/sec to a Seagate LTO tape drive over fibre channel. =
Is that the best I can hope for? And why wouldn't I see all the delays =
in "bptm" (if the tape drive is the bottleneck) or all in "bpbkar" (if =
the disk is the bottleneck)?
This is a NetBackup 4.5 media server, running Windows 2000 SP2, with =
plenty of memory and CPU capacity.
Thanks,
-Michael
------_=_NextPart_001_01C268BB.5E564C44
Content-Type: text/html;
charset="iso-8859-1"
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=3Diso-8859-1">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
6.0.6249.1">
<TITLE>Where's the bottleneck?</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->
<BR>
<P><FONT SIZE=3D2 FACE=3D"Arial">I've been trying to tune backups =
running on a particular media server in our environment and am not =
making much progress. I've modified the SIZE_DATA_BUFFERS and =
NUMBER_DATA_BUFFERS parameters (with local/SAN attached tape drives) but =
it doesn't seem to make much difference. From my last test, I saw =
the following messages in the "bpbkar" log file:</FONT></P>
<P><FONT SIZE=3D2 FACE=3D"Arial">3:13:41.250 PM: [2780.2308] <4> =
tar_backup::OVPC_EOFSharedMemory: INF - bpbkar waited 4198 times for =
empty buffer, delayed 6071 times</FONT></P>
<P><FONT SIZE=3D2 FACE=3D"Arial">That seems to imply that I don't have =
enough buffers, but from the "bptm" log file I see =
this:</FONT>
</P>
<P><FONT SIZE=3D2 FACE=3D"Arial">15:13:41.343 [1976.1984] <2> =
write_data: waited for full buffer 1608 times, delayed 1951 times</FONT>
</P>
<P><FONT SIZE=3D2 FACE=3D"Arial">which implies that "bpbkar" =
couldn't fill the buffers fast enough. So which is it? In =
this case, I was doing a very simple, best case performance test, =
backing up a single 4.2GB data file. I'm consistently getting =
about 13MB/sec to a Seagate LTO tape drive over fibre channel. Is =
that the best I can hope for? And why wouldn't I see all the =
delays in "bptm" (if the tape drive is the bottleneck) or all =
in "bpbkar" (if the disk is the bottleneck)?</FONT></P>
<P><FONT SIZE=3D2 FACE=3D"Arial">This is a NetBackup 4.5 media server, =
running Windows 2000 SP2, with plenty of memory and CPU capacity.</FONT>
</P>
<P><FONT SIZE=3D2 FACE=3D"Arial">Thanks,</FONT>
<BR><FONT SIZE=3D2 FACE=3D"Arial">-Michael</FONT>
</P>
</BODY>
</HTML>
------_=_NextPart_001_01C268BB.5E564C44--
|