[prev in list] [next in list] [prev in thread] [next in thread] 

List:       quagga-users
Subject:    [quagga-users 7147] SV: SV: Linux per-flow vs per packet load
From:       "Thomas Rasmussen" <tr () nmnet ! dk>
Date:       2006-06-29 6:56:33
Message-ID: EB5388B5BF49C943A7AA479116DF172F0643DD () exchange ! nmnet ! dk
[Download RAW message or body]

[Attachment #2 (multipart/alternative)]


To test the setup i used : HYPERLINK \
"http://www.netperf.org/netperf/NetperfPage.html"http://www.netperf.org/netperf/NetperfPage.html
  
 
First and most importent use :
 
Some intel based netcard : intel pci-x 1000MT
 
Turn on Freebsd fastwarding ( freebsd got a fastforwarding plane)
 
then recompiled the kernel with polling and tweaked the kernel.
 
 
thats it
 
Thomas 

   _____  

Fra: Michael Bernstein [mailto:mb_jobs@yahoo.com] 
Sendt: 29. juni 2006 00:01
Til: Thomas Rasmussen; quagga-users@lists.quagga.net
Emne: Re: SV: [quagga-users 7126] Linux per-flow vs per packet load balancing


Hi Thomas,
This is very interesting news. I was interested in what method you are using to \
determine throughput on the BSD router? Are you using any particular manufacturer of \
gbit nics? Are you tweaking the kernel or any specific parameters. Are you just using \
two interfaces?

Thanks for your help. 

Mike Bernstein

Thomas Rasmussen <tr@nmnet.dk> wrote: 

first thing i dont know anything about ospf ecmp or teql.. but one thing i know is \
iam able to move MORE packets per sec. then linux on freebsd..  Freebsd has a \
fastforwarding plane that just rocks.. thats why solaris is trying it implement this \
way of doing software routing..  Right now iam moving gb wirespeed with my freebsd \
routers..  
regards
Thomas Rasmussen

   _____  

Fra: quagga-users-bounces@lists.quagga.net \
                [mailto:quagga-users-bounces@lists.quagga.net] P� vegne af Michael \
                Bernstein
Sendt: 24. juni 2006 00:55
Til: quagga-users@lists.quagga.net
Emne: [quagga-users 7126] Linux per-flow vs per packet load balancing


Hi group,

My goal is per-packet load-balancing, not per-flow.  I wanted to explore OSPF ECMP, \
but I don't believe it can do per-packet. TEQL I heard is bad whether or not it can \
do per-packet or not. The "Equalize" keyword or flag setting I heard is bad news.

I'm looking at a Linux platform for this, or BSD, but BSD according to Gobble is NOT \
a good routing platform.

L2 channel bonding NICs may be an option, but what use would it have? I use bond0 on \
Sourcefire IDS's for work so I am familiar with the workings.

My goal is to achieve highest network performance using Linux routers for a fraction \
of the cost of a Cisco.

Any suggestions appreciated.

Regards,

Michael Bernstein
CCIE Security, CCNP, JNCIS, et. al.


   _____  

Do you Yahoo!?
Everyone is raving about the HYPERLINK \
"http://us.rd.yahoo.com/evt=40791/*http://advision.webevents.yahoo.com/handraisers"all-new \
Yahoo! Mail Beta.

--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.9.2/373 - Release Date: 22-06-2006



--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.9.3/374 - Release Date: 23-06-2006




   _____  

Do you Yahoo!?
Everyone is raving about the HYPERLINK \
"http://us.rd.yahoo.com/evt=40791/*http://advision.webevents.yahoo.com/handraisers"all-new \
Yahoo! Mail Beta.


--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.9.5/377 - Release Date: 27-06-2006



-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.9.5/377 - Release Date: 27-06-2006
 


[Attachment #5 (text/html)]

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">


<META content="MSHTML 6.00.2900.2912" name=GENERATOR></HEAD>
<BODY>
<DIV dir=ltr align=left>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2><SPAN class=038363506-29062006>To test the setup i used : 
<A 
href="http://www.netperf.org/netperf/NetperfPage.html">http://www.netperf.org/netperf/NetperfPage.html</A></SPAN></FONT></SPAN></DIV>
 <DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2><SPAN 
class=038363506-29062006></SPAN></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial><FONT 
color=#0000ff><FONT size=2>First and most importent<SPAN 
class=038363506-29062006> use :</SPAN></FONT></FONT></FONT></SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2><SPAN class=038363506-29062006>Some intel based netcard : 
intel pci-x 1000MT</SPAN></FONT></SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2><SPAN class=038363506-29062006>Turn on Freebsd fastwarding 
( freebsd got a fastforwarding plane)</SPAN></FONT></SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial 
color=#0000ff size=2></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial><FONT 
color=#0000ff><FONT size=2>then&nbsp;<SPAN class=038363506-29062006>recompiled 
the kernel with polling and tweaked the 
kernel.</SPAN></FONT></FONT></FONT></SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial><FONT 
color=#0000ff><FONT size=2><SPAN 
class=038363506-29062006></SPAN></FONT></FONT></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial><FONT 
color=#0000ff><FONT size=2><SPAN 
class=038363506-29062006></SPAN></FONT></FONT></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial><FONT 
color=#0000ff><FONT size=2><SPAN class=038363506-29062006>thats 
it</SPAN></FONT></FONT></FONT></SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial><FONT 
color=#0000ff><FONT size=2><SPAN 
class=038363506-29062006></SPAN></FONT></FONT></FONT></SPAN>&nbsp;</DIV>
<DIV dir=ltr align=left><SPAN class=535453406-29062006><FONT face=Arial><FONT 
color=#0000ff><FONT size=2><SPAN class=038363506-29062006>Thomas 
</SPAN></FONT></FONT></FONT></SPAN></DIV></DIV><BR>
<DIV class=OutlookMessageHeader lang=da dir=ltr align=left>
<HR tabIndex=-1>
<FONT face=Tahoma size=2><B>Fra:</B> Michael Bernstein 
[mailto:mb_jobs@yahoo.com] <BR><B>Sendt:</B> 29. juni 2006 00:01<BR><B>Til:</B> 
Thomas Rasmussen; quagga-users@lists.quagga.net<BR><B>Emne:</B> Re: SV: 
[quagga-users 7126] Linux per-flow vs per packet load 
balancing<BR></FONT><BR></DIV>
<DIV></DIV>Hi Thomas,<BR>This is very interesting news. I was interested in what 
method you are using to determine throughput on the BSD router? Are you using 
any particular manufacturer of gbit nics? Are you tweaking the kernel or any 
specific parameters. Are you just using two interfaces?<BR><BR>Thanks for your 
help. <BR><BR>Mike Bernstein<BR><BR><B><I>Thomas Rasmussen 
&lt;tr@nmnet.dk&gt;</I></B> wrote:
<BLOCKQUOTE class=replbq 
style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: rgb(16,16,255) 2px solid">
  <META content="MSHTML 6.00.2900.2912" name=GENERATOR>
  <DIV dir=ltr align=left><SPAN class=187541809-24062006><FONT face=Arial 
  color=#0000ff size=2>first thing i dont know anything about ospf ecmp or 
  teql.. but one thing i know is iam able to move MORE packets per sec. then 
  linux on freebsd..&nbsp; Freebsd has a fastforwarding plane that just rocks.. 
  thats why solaris is trying it implement this way of doing software 
  routing..&nbsp; Right now iam moving gb wirespeed with my freebsd 
  routers..</FONT></SPAN></DIV>
  <DIV dir=ltr align=left><SPAN class=187541809-24062006><FONT face=Arial 
  color=#0000ff size=2></FONT></SPAN>&nbsp;</DIV>
  <DIV dir=ltr align=left><SPAN class=187541809-24062006><FONT face=Arial 
  color=#0000ff size=2>regards</FONT></SPAN></DIV>
  <DIV dir=ltr align=left><SPAN class=187541809-24062006><FONT face=Arial 
  color=#0000ff size=2>Thomas Rasmussen</FONT></SPAN></DIV><BR>
  <DIV class=OutlookMessageHeader lang=da dir=ltr align=left>
  <HR tabIndex=-1>
  <FONT face=Tahoma size=2><B>Fra:</B> quagga-users-bounces@lists.quagga.net 
  [mailto:quagga-users-bounces@lists.quagga.net] <B>P� vegne af </B>Michael 
  Bernstein<BR><B>Sendt:</B> 24. juni 2006 00:55<BR><B>Til:</B> 
  quagga-users@lists.quagga.net<BR><B>Emne:</B> [quagga-users 7126] Linux 
  per-flow vs per packet load balancing<BR></FONT><BR></DIV>
  <DIV></DIV>Hi group,<BR><BR>My goal is per-packet load-balancing, not 
  per-flow.&nbsp; I wanted to explore OSPF ECMP, but I don't believe it can do 
  per-packet. TEQL I heard is bad whether or not it can do per-packet or not. 
  The "Equalize" keyword or flag setting I heard is bad news.<BR><BR>I'm looking 
  at a Linux platform for this, or BSD, but BSD according to Gobble is NOT a 
  good routing platform.<BR><BR>L2 channel bonding NICs may be an option, but 
  what use would it have? I use bond0 on Sourcefire IDS's for work so I am 
  familiar with the workings.<BR><BR>My goal is to achieve highest network 
  performance using Linux routers for a fraction of the cost of a 
  Cisco.<BR><BR>Any suggestions appreciated.<BR><BR>Regards,<BR><BR>Michael 
  Bernstein<BR>CCIE Security, CCNP, JNCIS, et. al.<BR><BR>
  <DIV></DIV>
  <HR SIZE=1>
  Do you Yahoo!?<BR>Everyone is raving about the <A 
  href="http://us.rd.yahoo.com/evt=40791/*http://advision.webevents.yahoo.com/handraisers">all-new \
  Yahoo! Mail Beta.</A><BR>
  <DIV><FONT size=2>--<BR>No virus found in this incoming message.<BR>Checked by 
  AVG Free Edition.<BR>Version: 7.1.394 / Virus Database: 268.9.2/373 - Release 
  Date: 22-06-2006<BR></FONT></DIV>
  <DIV><FONT face=Arial size=2></FONT></DIV><BR>
  <DIV><FONT size=2>--<BR>No virus found in this outgoing message.<BR>Checked by 
  AVG Free Edition.<BR>Version: 7.1.394 / Virus Database: 268.9.3/374 - Release 
  Date: 23-06-2006<BR></FONT></DIV></BLOCKQUOTE><BR>
<P>
<HR SIZE=1>
Do you Yahoo!?<BR>Everyone is raving about the <A 
href="http://us.rd.yahoo.com/evt=40791/*http://advision.webevents.yahoo.com/handraisers">all-new \
 Yahoo! Mail Beta.</A><BR>
<P><FONT size=2>--<BR>No virus found in this incoming message.<BR>Checked by AVG 
Free Edition.<BR>Version: 7.1.394 / Virus Database: 268.9.5/377 - Release Date: 
27-06-2006<BR></FONT></P></BODY></HTML>
<BR>

<P><FONT SIZE=2>--<BR>
No virus found in this outgoing message.<BR>
Checked by AVG Free Edition.<BR>
Version: 7.1.394 / Virus Database: 268.9.5/377 - Release Date: 27-06-2006<BR>
</FONT> </P>



_______________________________________________
Quagga-users mailing list
Quagga-users@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-users


[prev in list] [next in list] [prev in thread] [next in thread] 

Configure | About | News | Add a list | Sponsored by KoreLogic