X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C8BCF5.DE3E14F8@onstor-exch02.onstor.net>; Fri, 23 May 2008 09:55:56 -0700
MIME-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C8BCF5.DE3E14F8"
Content-class: urn:content-classes:message
Subject: RE: Cougar and Bobcat baseline performance testing with LSI 3992 SATA storage
Date: Fri, 23 May 2008 09:55:56 -0700
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E0A1C26F2@onstor-exch02.onstor.net>
In-Reply-To: <BB375AF679D4A34E9CA8DFA650E2B04E0A1C26D7@onstor-exch02.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Cougar and Bobcat baseline performance testing with LSI 3992 SATA storage
Thread-Index: Aci66/jGYRoDe3BDSTKu+8qrl4afCwBTQ3YwAC3ElDAAAOhpIAAAOEWw
References: <BB375AF679D4A34E9CA8DFA650E2B04E0A1C2694@onstor-exch02.onstor.net> <BB375AF679D4A34E9CA8DFA650E2B04E0A1C26D7@onstor-exch02.onstor.net>
From: "Brian Stark" <brian.stark@onstor.com>
To: "Fay Chong" <fay.chong@onstor.com>,
	"Maxim Kozlovsky" <maxim.kozlovsky@onstor.com>,
	"dl-Cougar" <dl-Cougar@onstor.com>
Cc: "dl-qualification" <dl-qualification@onstor.com>

This is a multi-part message in MIME format.

------_=_NextPart_001_01C8BCF5.DE3E14F8
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

The next step will be to implement the hardware checksum feature that is =
supported in the 1480.  I will discuss this with some of the Cougar dev =
folks.


Brian


> _____________________________________________=20
> From: 	Fay Chong =20
> Sent:	Friday, May 23, 2008 9:43 AM
> To:	Maxim Kozlovsky; Brian Stark; dl-Cougar
> Cc:	dl-qualification
> Subject:	RE: Cougar and Bobcat baseline performance testing with LSI =
3992 SATA storage
>=20
> Thanks for the analysis Max.=20
>=20
> Cougar team, what> '> s the next step? How can I help?
>=20
> Thanks
> Fay
>=20
> _____________________________________________
> From: Maxim Kozlovsky=20
> Sent: Friday, May 23, 2008 9:15 AM
> To: Brian Stark; Fay Chong; dl-Cougar
> Cc: dl-qualification
> Subject: RE: Cougar and Bobcat baseline performance testing with LSI =
3992 SATA storage
>=20
> It is because of software checksums. Ncpu is out of cpu.
>=20
> _____________________________________________
> From: Brian Stark=20
> Sent: Thursday, May 22, 2008 9:08 PM
> To: Fay Chong; dl-Cougar
> Cc: dl-qualification
> Subject: RE: Cougar and Bobcat baseline performance testing with LSI =
3992 SATA storage
>=20
> We've shown with nfspertest that the bandwidth inside the box is much =
higher than Bobcat, but this doesn't take into account bandwidth over =
the GigE links.  The only explanation that I can think of for the =
difference is that Cougar is currently doing TCP checksums in software, =
which is causing the slowdown in performance relative to Bobcat.  We =
didn't see a bottleneck doing the software checksums when running SPEC, =
but it's possible that streaming workloads are affected.
>=20
> Can we get a profile on the TXRX when running this test?  This would =
then show if the processor is spending a lot of time in the checksum =
code.
>=20
>=20
> Brian
>=20
>=20
> 	_____________________________________________=20
> 	From: 	Fay Chong =20
> 	Sent:	Tuesday, May 20, 2008 7:40 PM
> 	To:	dl-Cougar
> 	Cc:	dl-qualification
> 	Subject:	Cougar and Bobcat baseline performance testing with LSI 3992 =
SATA storage
>=20
> 	Here are the results of Cougar and Bobcat baseline testing with LSI =
3992 SATA storage. The Cougar was slightly better in sequential writes =
but worse in sequential reads. Random reads and writes were about the =
same. A front end FP bandwidth test was performed. This test uses =
iometer with =BD GB files on 4 raid groups. Sequential reads were =
executed (32k). It was expected that the entire files would be in the =
filer cache so this should be a measure of front end read bandwidth. =
Here are the results:
>=20
> 	Cougar (R4.0 sub 22) =3D 145 MB/sec=20
> 	Bobcat (R3.3 sub 22) =3D 221 MB/sec=20
>=20
> 	One would have expected the Cougar to have greater bandwidth than the =
Bobcat. More investigation should be done.
>=20
> 	=
\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Cougar-LSI3992pub.xl=
s
>=20
> 	=
\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Bobcat2280-LSI3992pu=
b.xls
>=20
> 	=
\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Cougar-Bobcatpub.xls=

>=20
> 	Thanks
> 	Fay
>=20
>=20
>=20
> 	Fay Chong
> 	Sr. Performance Engineer
> 	ONStor, Inc.
> 	fay.chong@onstor.com
> 	408.376.3130 (w)
>=20
>=20

------_=_NextPart_001_01C8BCF5.DE3E14F8
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 =
6.5.7653.38">
<TITLE>RE: Cougar and Bobcat baseline performance testing with LSI 3992 =
SATA storage</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->

<P><FONT COLOR=3D"#0000FF" SIZE=3D2 FACE=3D"Arial">The next step will be =
to implement the hardware checksum feature that is supported in the =
1480.&nbsp; I will discuss this with some of the Cougar dev =
folks.</FONT></P>
<BR>

<P><FONT COLOR=3D"#0000FF" SIZE=3D2 FACE=3D"Arial">Brian</FONT>
</P>
<BR>
<UL>
<P><FONT SIZE=3D1 =
FACE=3D"Tahoma">_____________________________________________ </FONT>

<BR><B><FONT SIZE=3D1 FACE=3D"Tahoma">From: &nbsp;</FONT></B> <FONT =
SIZE=3D1 FACE=3D"Tahoma">Fay Chong&nbsp; </FONT>

<BR><B><FONT SIZE=3D1 FACE=3D"Tahoma">Sent:&nbsp;&nbsp;</FONT></B> <FONT =
SIZE=3D1 FACE=3D"Tahoma">Friday, May 23, 2008 9:43 AM</FONT>

<BR><B><FONT SIZE=3D1 =
FACE=3D"Tahoma">To:&nbsp;&nbsp;&nbsp;&nbsp;</FONT></B> <FONT SIZE=3D1 =
FACE=3D"Tahoma">Maxim Kozlovsky; Brian Stark; dl-Cougar</FONT>

<BR><B><FONT SIZE=3D1 =
FACE=3D"Tahoma">Cc:&nbsp;&nbsp;&nbsp;&nbsp;</FONT></B> <FONT SIZE=3D1 =
FACE=3D"Tahoma">dl-qualification</FONT>

<BR><B><FONT SIZE=3D1 =
FACE=3D"Tahoma">Subject:&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</FONT>=
</B> <FONT SIZE=3D1 FACE=3D"Tahoma">RE: Cougar and Bobcat baseline =
performance testing with LSI 3992 SATA storage</FONT>
</P>

<P><FONT COLOR=3D"#000080" SIZE=3D2 FACE=3D"Arial">Thanks for the =
analysis Max. </FONT>
</P>

<P><FONT COLOR=3D"#000080" SIZE=3D2 FACE=3D"Arial">Cougar team, =
what&#8217;s the next step? How can I help?</FONT>
</P>

<P><FONT COLOR=3D"#000080" SIZE=3D2 FACE=3D"Arial">Thanks</FONT>

<BR><FONT COLOR=3D"#000080" SIZE=3D2 FACE=3D"Arial">Fay</FONT>
</P>

<P><FONT SIZE=3D2 =
FACE=3D"Tahoma">_____________________________________________<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">From:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Maxim Kozlovsky<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">Sent:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Friday, May 23, 2008 9:15 AM<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">To:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Brian Stark; Fay Chong; dl-Cougar<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">Cc:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> dl-qualification<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">Subject:</FONT></B><FONT =
SIZE=3D2 FACE=3D"Tahoma"> RE: Cougar and Bobcat baseline performance =
testing with LSI 3992 SATA storage</FONT>
</P>

<P><FONT COLOR=3D"#000080" SIZE=3D2 FACE=3D"Arial">It is because of =
software checksums. Ncpu is out of cpu.</FONT>
</P>

<P><FONT SIZE=3D2 =
FACE=3D"Tahoma">_____________________________________________<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">From:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Brian Stark<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">Sent:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Thursday, May 22, 2008 9:08 PM<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">To:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Fay Chong; dl-Cougar<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">Cc:</FONT></B><FONT SIZE=3D2 =
FACE=3D"Tahoma"> dl-qualification<BR>
</FONT><B><FONT SIZE=3D2 FACE=3D"Tahoma">Subject:</FONT></B><FONT =
SIZE=3D2 FACE=3D"Tahoma"> RE: Cougar and Bobcat baseline performance =
testing with LSI 3992 SATA storage</FONT>
</P>

<P><FONT COLOR=3D"#0000FF" SIZE=3D2 FACE=3D"Arial">We've shown with =
nfspertest that the bandwidth inside the box is much higher than Bobcat, =
but this doesn't take into account bandwidth over the GigE links.&nbsp; =
The only explanation that I can think of for the difference is that =
Cougar is currently doing TCP checksums in software, which is causing =
the slowdown in performance relative to Bobcat.&nbsp; We didn't see a =
bottleneck doing the software checksums when running SPEC, but it's =
possible that streaming workloads are affected.</FONT></P>

<P><FONT COLOR=3D"#0000FF" SIZE=3D2 FACE=3D"Arial">Can we get a profile =
on the TXRX when running this test?&nbsp; This would then show if the =
processor is spending a lot of time in the checksum code.</FONT></P>
<BR>

<P><FONT COLOR=3D"#0000FF" SIZE=3D2 FACE=3D"Arial">Brian</FONT>
</P>
<BR>
<UL>
<P><FONT SIZE=3D1 =
FACE=3D"Tahoma">_____________________________________________ </FONT>

<BR><B><FONT SIZE=3D1 FACE=3D"Tahoma">From: &nbsp;</FONT></B> <FONT =
SIZE=3D1 FACE=3D"Tahoma">Fay Chong&nbsp; </FONT>

<BR><B><FONT SIZE=3D1 FACE=3D"Tahoma">Sent:&nbsp;&nbsp;</FONT></B> <FONT =
SIZE=3D1 FACE=3D"Tahoma">Tuesday, May 20, 2008 7:40 PM</FONT>

<BR><B><FONT SIZE=3D1 =
FACE=3D"Tahoma">To:&nbsp;&nbsp;&nbsp;&nbsp;</FONT></B> <FONT SIZE=3D1 =
FACE=3D"Tahoma">dl-Cougar</FONT>

<BR><B><FONT SIZE=3D1 =
FACE=3D"Tahoma">Cc:&nbsp;&nbsp;&nbsp;&nbsp;</FONT></B> <FONT SIZE=3D1 =
FACE=3D"Tahoma">dl-qualification</FONT>

<BR><B><FONT SIZE=3D1 =
FACE=3D"Tahoma">Subject:&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</FONT>=
</B> <FONT SIZE=3D1 FACE=3D"Tahoma">Cougar and Bobcat baseline =
performance testing with LSI 3992 SATA storage</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Here are the results of Cougar and =
Bobcat baseline testing with LSI 3992 SATA storage. The Cougar was =
slightly better in sequential writes but worse in sequential reads. =
Random reads and writes were about the same. A front end FP bandwidth =
test was performed. This test uses iometer with =BD GB files on 4 raid =
groups. Sequential reads were executed (32k). It was expected that the =
entire files would be in the filer cache so this should be a measure of =
front end read bandwidth. Here are the results:</FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Cougar (R4.0 sub 22) =3D 145 MB/sec =
</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Arial">Bobcat (R3.3 sub 22) =3D 221 MB/sec =
</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">One would have expected the Cougar to =
have greater bandwidth than the Bobcat. More investigation should be =
done.</FONT>
</P>

<P><A =
HREF=3D"file://\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Couga=
r-LSI3992pub.xls"><U><FONT COLOR=3D"#0000FF" SIZE=3D2 =
FACE=3D"Arial">\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Couga=
r-LSI3992pub.xls</FONT></U></A>
</P>

<P><A =
HREF=3D"file://\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Bobca=
t2280-LSI3992pub.xls"><U><FONT COLOR=3D"#0000FF" SIZE=3D2 =
FACE=3D"Arial">\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Bobca=
t2280-LSI3992pub.xls</FONT></U></A>
</P>

<P><A =
HREF=3D"file://\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Couga=
r-Bobcatpub.xls"><U><FONT COLOR=3D"#0000FF" SIZE=3D2 =
FACE=3D"Arial">\\Mightydog\users\fayc\public\WorkInProgress\LSI3992\Couga=
r-Bobcatpub.xls</FONT></U></A>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Thanks</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Arial">Fay</FONT>
</P>
<BR>
<BR>

<P><FONT SIZE=3D2 FACE=3D"Arial">Fay Chong</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Arial">Sr. Performance Engineer</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Arial">ONStor, Inc.</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Arial">fay.chong@onstor.com</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Arial">408.376.3130 (w)</FONT>
</P>
<BR>
</UL></UL>
</BODY>
</HTML>
------_=_NextPart_001_01C8BCF5.DE3E14F8--
