Mailing List Archive

Use Supersparrow http://supersparrow.sourceforge.net with LVS and you get
the same thing...

They are in the same space as radware, foundry serveriron, arrowpoint (now
cisco), F5 etc...

You can find some of the load balancing sites listed here:
http://dmoz.org/Computers/Software/Internet/Site_Management/Load_Balancing/
-----Original Message-----
From: lvs-users-admin@LinuxVirtualServer.org
[mailto:lvs-users-admin@LinuxVirtualServer.org]On Behalf Of Joseph Mack
Sent: Monday, March 19, 2001 9:49 AM
To: lvs-users@LinuxVirtualServer.org
Subject: Coyote Point Load Balancer


Anyone know anything about this? An ad flashed up
on my screen on Slashdot about it

http://www.coyotepoint.com/

It appears to have much the same functionality as LVS
as well as having geographically distributed load balancing.

Joe

--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA

_______________________________________________
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to lvs-users-request@LinuxVirtualServer.org
or go to http://www.in-addr.de/mailman/listinfo/lvs-users

I took a quick look.

Seems like a low-end solution. (8000000 connections per hour
works out to 2k connections per second).

/sG

LVS Account wrote:
>
> Use Supersparrow http://supersparrow.sourceforge.net with LVS and you get
> the same thing...

I'm amazed. This is Horms code that I've been waiting to see for quite a while.
Didn't even know it had come out.

> You can find some of the load balancing sites listed here:
> http://dmoz.org/Computers/Software/Internet/Site_Management/Load_Balancing/

thanks. Didn't know about this either.

Joe
--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA

Steve Gonczi wrote:
>
> I took a quick look.
>
> Seems like a low-end solution. (8000000 connections per hour
> works out to 2k connections per second).

It's US$5k for device that can handle a T1 line. Seems a lot of money to me.

Joe

--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@epa.gov ph# 919-541-0007, RTP, NC, USA

The current (probably April 2001) issue of Linux Journal has an article from
a pair of Erikson engineers who were looking at using LVS to front end a
bunch of diskless web servers. They only tried NAT in the article, and on a
500mhz pIII got about 1700-1800 connections per second before it topped out
(tested using Web Bench). Direct connection to the servers yielded around
8000 connects/sec. What wasn't clear was where the bottleneck was. Would a
Ghz class pIII double the throughput, would better NIC's help, is there a
limitation of the Linux stack? They were going to try the DR mode, but
hadn't by press time. I believe they were using the stock 6.2 RedHat
Piranha dist. The 2.4 kernel is reputed to have a more efficient network
stack, does anyone have a feel for the amount of perf improvement for LVS
(understanding that Web Bench is a benchmark and nothing more)?

jim

-----Original Message-----
From: Steve Gonczi [mailto:Steve.Gonczi@networkengines.com]
Sent: Monday, March 19, 2001 9:01 AM
To: 'lvs-users@LinuxVirtualServer.org'
Subject: RE: Coyote Point Load Balancer


I took a quick look.

Seems like a low-end solution. (8000000 connections per hour
works out to 2k connections per second).

/sG

_______________________________________________
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to lvs-users-request@LinuxVirtualServer.org
or go to http://www.in-addr.de/mailman/listinfo/lvs-users

Jim,

In a NAT configuration, the primary bottleneck is the return traffic.
Web traffic tends to be very asymmetrical (the ratio between incoming and
outgoing traffic may be as high as 1:50 and the front-end server has to
process all return traffic).

With DR, the bottleneck will be the incoming NIC. (Probably the
number of packets the driver can receive). I expect this to be
as high as 10-12K requests per second on a 100Mb line, with
a reasonable NIC/driver. DR has some limitations,
but it is clearly the fastest way to go.

The stack will not make a big difference, as most LVS processing
happens in the kernel, just above the driver layer.

/sG


-----Original Message-----
From: Feldman, Jim [mailto:Jim.Feldman@COMPAQ.com]
Sent: Monday, March 19, 2001 2:39 PM
To: 'lvs-users@LinuxVirtualServer.org'
Subject: RE: Coyote Point Load Balancer


The current (probably April 2001) issue of Linux Journal has an article from
a pair of Erikson engineers who were looking at using LVS to front end a
bunch of diskless web servers. They only tried NAT in the article, and on a
500mhz pIII got about 1700-1800 connections per second before it topped out
(tested using Web Bench). Direct connection to the servers yielded around
8000 connects/sec. What wasn't clear was where the bottleneck was. Would a
Ghz class pIII double the throughput, would better NIC's help, is there a
limitation of the Linux stack? They were going to try the DR mode, but
hadn't by press time. I believe they were using the stock 6.2 RedHat
Piranha dist. The 2.4 kernel is reputed to have a more efficient network
stack, does anyone have a feel for the amount of perf improvement for LVS
(understanding that Web Bench is a benchmark and nothing more)?

jim

-----Original Message-----
From: Steve Gonczi [mailto:Steve.Gonczi@networkengines.com]
Sent: Monday, March 19, 2001 9:01 AM
To: 'lvs-users@LinuxVirtualServer.org'
Subject: RE: Coyote Point Load Balancer


I took a quick look.

Seems like a low-end solution. (8000000 connections per hour
works out to 2k connections per second).

/sG

_______________________________________________
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to lvs-users-request@LinuxVirtualServer.org
or go to http://www.in-addr.de/mailman/listinfo/lvs-users

_______________________________________________
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to lvs-users-request@LinuxVirtualServer.org
or go to http://www.in-addr.de/mailman/listinfo/lvs-users

On Mon, 19 Mar 2001, Steve Gonczi wrote:

> Jim,
>
> In a NAT configuration, the primary bottleneck is the return traffic.
> Web traffic tends to be very asymmetrical (the ratio between incoming and
> outgoing traffic may be as high as 1:50 and the front-end server has to
> process all return traffic).

We all thought this till someone went and measured what was happening.
This assymmetry in byte throughput turns out not to matter. LVS is a
packet pusher and it's the number of packets that are important.

http://www.linuxvirtualserver.org/Joseph.Mack/HOWTO/LVS-HOWTO_1.0-5.html#ss5.2

Joe

--
Joseph Mack mack@ncifcrf.gov

>This assymmetry in byte throughput turns out not to matter.
>LVS is a
>packet pusher and it's the number of packets that are important.

Coming in, you are absolutely right. You will reach the max packet rate
bottleneck, way before you would saturate the incoming NIC.

I believe that you will saturate the outgoing bandwidth, before you reach
the
incoming max packet rate, unless you do DR. (Assuming you are serving
static web pages. Streaming media, or heavy cgi processing is a different
story).

BTW the link
http://www.linuxvirtualserver.org/Joseph.Mack/performance/single_real-server
_performance.html appears to be broken.

cheers,

/sG