|
On Wed, 24 Sep 2003, Horms wrote:
> Hi,
>
> I am looking for information from people who are running LVS in
> production. I am most interested in how reliable they find
> LVS to be.
Extremely reliable, provided you run the LVS routers on good hardware, and
maintain the OS w/ patches as needed.
> How often does the system go down.
In 3 years, never. We run a pair of LVS Routers and several cluster nodes
and we routinely do maintenance on them. We can lose 2/3 of our
infrastructure and still function, but the most we ever do is take on box
out of service at a time to update and patch. The LVS Routers simply pull
it out of the cluster and put it back in when it is ready.
> And when it does what is the cause - if it is caused by something
> unrelated to LVS like faulty hardware then you can skip that bit.
During testing phases it was due to idiotic mistakes that we made in
configuration and a lack of understanding of how to manage the platform.
In production, it hasn't been an issue. We've scaled the cluster up to
1,000 Web hits a second in testing (static content, large images) and it
completely kicks the crap out of everything else we've seen (F5, Cisco
etc..)
> If anyone is able to volunteer this information either
> on the list or to me privately then I would be most grateful.
> Please include some rough information on your system:
> What architecture, which kernel, which version of LVS,
> approximately how many nodes, and how busy it is.
We started with custom kernels but have migrated our LVS routers to RedHat
Advanced Server, running on $800 1U Celeron based servers. The Cluster
nodes are $1,200 1U Celeron servers. Currently, the system runs with 8
total machines (6 nodes, 2 directors). The system is lighlty loaded and
provides Load Balancing for SMTP, FTP, POP3, HTTP and HTTPS services. By
design, we have each cluster node doing Webmail, Pop3, SMTP, Virus/Spam
Stripping, Ftp and Web services. We make Webmail and HTTPS requests sticky
accross the cluster. When new viruses are released, the system takes the
load and laughs at it.
> I plan to use this information as part of a presentation
> that I will be making on LVS at Linux.Conf.Au[1] as well
> as when trying to convince potential customers[2] to use
> LVS. If I get enough information to make this interesting
> I will put it up on a web page somewhere.
Hope this helps! We lease access to our LVS cluster to our clients who
need a solution that scales. Mostly these are people that do national
advertising campaigns that hit on a specific day at a specific hour, and
the load on the clusters shoots up. The great thing is that we can do
scalability things with the cluster w/ MS backend nodes that MS can't even
do itself, and when we spec out a solution on our cluster, it usually
comes out at 1/2 the price of the competition at about 2x the
profitability.
It's a completely enterprise ready solution and I'd advise everyone to use
it!
> [1] http://www.linux.org.au/conf/
> [2] As many of you may know I work for VA Linux Systems Japan
> http://www.valinux.co.jp/ This is no longer the same as
> VA Linux Systems USA where I previously worked! Part of
> what we do is to sell system integration solutions that include LVS.
>
> --
> Horms
> _______________________________________________
> LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
> or go to http://www.in-addr.de/mailman/listinfo/lvs-users
>
|
