LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Oddity building LBLC

To: Thomas Proell <Thomas.Proell@xxxxxxxxxx>
Subject: Re: Oddity building LBLC
Cc: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
From: Martin Hamilton <martin@xxxxxxxxxxxxx>
Date: Wed, 15 Nov 2000 13:22:55 +0000
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Content-Type: text/plain; charset=us-ascii

Thomas Proell writes:

| Do you use the Redhat-Kernel or the original Linux Kernel?
| Redhat may not work :-(

This is the real kosher 2.2.17 kernel - I was surprised that it 
didn't work for me, since nobody else on the list appeared to have a 
problem.  Have many people been playing with LBLC yet ?  Any 
experiences to share ?

FWIW I should mention that we (see http://wwwcache.ja.net/) are very
interested in something like this because we currently use LVS to load
balance some 80 million URL requests in three Squid proxy cache server
farms.  At the moment we're internally peering our caches with each
other using the Squid cache digest mechanism (ICP was too expensive),
but would love to retire this in favour of a hashing type system at 
the director end - if it turned out to be practical.

Our initial thought was to see how we could expect LBLC to perform by
putting a test front end up and replaying requests from our logs,
using a single (test) real server which peered with our genuine real
servers, and (say) fetching only cache hits to avoid generating lots
of real world traffic/transactions.  May get a chance at this later 
on today...

Cheers,

Martin




-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.4 (GNU/Linux)
Comment: Exmh version 2.1.1 10/15/1999 + martin

iD8DBQE6Eo4uVw+hz3xBJfQRAsOmAJ9LKPFf4NnrYLESmCaKrj8C202DvgCguceY
V5P8nhAX+w/t0aAQ3IuVAFU=
=MFvS
-----END PGP SIGNATURE-----



<Prev in Thread] Current Thread [Next in Thread>