Jan-Frode Myklebust a écrit :
On 2009-03-17, John Lauro <[email protected]> wrote:
You need to explain a little more, as I am not understating something.
Perhaps what you mean by VIP?
Virtual IP address. With heartbeat, one normally has one staticly defined
ip-address on the frontend interface on each server, and then additionally
one or more VIPs that can be moved between servers.
Server1 has static IP on eth0 and the VIP eth0:1
Server2 has static IP on eth0
If server1 fails, server2 takes over eth0:1.
If they share the same single VIP at the same time, then why would you use
round-robin DNS? Round-robin is for multiple IP addresses...?
I use one VIP on each server, and use round robin DNS to distribute the
load over all the servers. If one of the servers go down / is taken down
for maintenance, I move its VIP to the other server.
Also, if you do a virtual IP like Microsoft Windows does for their multicast
load balancing, that is just plain nasty to your network infrastructure if
you have more than those servers on the same subnet and IMHO really doesn't
scale well...
That's true that Micro$oft use a lot of bandwith, but keepalived and
haertbeat dont ! They generated a low traffic impact. You can also
specify the frequence on multicast which can be 1 / sec if you want.
Also if your worry about multicast you can specify on which nic the
multicast will happen. What we usually do is connect a cross cable
between server1 and server2 on a seperate nic and you we tell keepalived
to use that seperate nic for cluster traffic, ( The traffic that will
tell what server are up, not the actual web traffic)
Cheer.
That doesn't sound like what we do, no.
If you meant a different VIP instead of one bound to each server, I could
understand that. However, 50% of the clients will feel the hit when first
connecting if a server is down.
Not when the VIP moves over to the server that's still up, which is what
heartbeat does for me.
-jf
--
Guillaume Bourque, B.Sc.,
consultant, infrastructures technologiques libres !
514 576-7638
