Need help planning implementation
 

Hey guys, so here is the deal. This is done entirely with cloud computing.

I have a load balancer with 2 web servers behind it. The web servers rsync with cloud storage to update their apache directories 1 time every hour. Apache is just running php pages that pull/push data to a DB so they dont need to be updated that often. However I need to figure out how to implement a Master/Master MySQL setup to have my web servers point to for the PHP stuff. I need to implement it without having a single point of failure. The Load balancers are useless for failover as they only detect availability based on Ping request. So putting a master/master setup behind a Load Balancer is out.

what is the best way to setup the master/master mysql in a HA setup without the use of a load balancer provided by the host?

Do you have your heart set on a multi-master cluster?

If not I'd recommend RHCS cluster with DRBD. Totally free with centos. Its simple to use and works.

multi-master clusters have their own issues with performance, maintenance and the hassle of mysql proxy if needed.

Loadbalancer would be the prefer method; not sure what load balancing tool you are using; but the following (FREE solution) works well and you can configure any kind of monitor you want: see: http://www.linuxvirtualserver.org/do...tbeat_mon.html. Even if you use a Load balancing tool you will want to create a script or something to tail your transaction log files and ensure any INSERTS|UPDATES|DELETES are replicated to the second mysql server unless you are using mysql clustering. Another method that I used years ago before LVS was any good; was Zebra; Creating an OSPF network between your web servers and your mysql servers and then you can make your mysql servers neighbors and weight the traffic and if the neighbor stops responding to OSPF updates then OSPF sends to next available route. kinda messy though now that LB's take the mess away.