Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 10 Nov 2002 14:24:49 +0100 (MET)
From:      Michael Grant <mg-fbsd3@grant.org>
To:        freebsd-cluster@freebsd.org
Subject:   Re: clustering freebsd
Message-ID:  <200211101324.gAADOng29727@splat.grant.org>

next in thread | raw e-mail | index | archive | help
Thanks, all of you, for your excellent replies.

After reading several replies, yes, it's load balancing that I want to 
do.  That looks very promising to solve the front-end machine problem.

I have started looking around for load balancer devices, there seem to 
be a quite a few on the market, they seem to be in the US$10K range.
Some of them like Cisco's LocalDirector product seem to work in
conjunction with a router.  It's possible that my ISP has one of
these or something similar.

A quite cool one is here, only 1U high and fully redundant:

http://www.loadbalancer.org/modules.php?name=Content&pa=showpage&pid=9

I'll look more into freevrrpd and loadd.

On the shared disk side, I've not found the perfect solution (yet).
This does seem to be a sticky problem.  Several of you have said I
don't want to do this.  Well, I think I do and worst case, I'll end up 
using a separate HA configured box and nfs mount it like Compaq's
prolient cluster thingy.  I haven't give up looking for a better
solution though.

The need for having a shared writable file system mainly comes from
having shell users with home directories and mail boxes.  I have had
great resistance from the users trying to move mailboxes into another
format other than plain mbox format.  i.e. moving mail into mysql
would not be a popular idea.

If mysql can replicate things like that, I wonder about implementing a 
file system ontop of mysql?  The performance would probably suck
though.  I'm surprised that there isn't some extension to JFS (the
Journaling File System) to do something like I want.

Alexander mentions AFS.  In fact, the folks who brought you AFS have
something called CODA which seems to be a network replicated file
system.  I read up on using it and it's quite complicated and seems to
have some restrictions on the size of file systems.  It also seems
like it's still in an experimental state.  It did't give me a warm
fuzzy feeling, but it certainly is cool.  Anyone else have experience
using it?

I couldn't find much on iSCSI for freebsd and it's not clear to me
that you could have n systems writing to a raid array with vinum.  I
can believe easily that you could have one system write to a vinum
raid array spread over several systems via iscsi though.

If anyone has any ideas on this front, I'd like to hear them.

Michael Grant

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-cluster" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200211101324.gAADOng29727>