I have set up successfully a high availability, load balanced MySQL cluster using the HowTo : How To Set Up A Load-Balanced MySQL Cluster from Falko. I had to adapt some of the commands because I'm using Debian Lenny but everything works like a charm.
The setup as I have it now is:
LoadBalancer1 with the Cluster management.
LoadBalancer2 failover config in case LoadBalancer1 goes down
I was wondering, since both LoadBalancers don't use a lot of resources, if I can run the same on only two servers, configuring the following:
MySQL1 being the cluster manager and first SQL node, with UltraMonkey for heartbeat and ldirector.
MySQL2 being the failover for management and the second SQL node, with UltraMonkey failover.
I'm not sure if one of the nodes goes down, what would be the consequence on the data level, will the cluster keep running on one machine and replicate the data when the other one comes up again? Or is this whole thing a bad idea?
This question of course was posed to me by my chief who is interested in limiting our expense
Now that I have it set up and running pretty good, he's coming up with this so any help to provide me with arguments in order to change his mind is welcome.
If someone with more experience could give me some pro's and con's of this change, I'd really appreciate it.