I've been tasked to explore ways of building a very fail-resistant configuration for a professional application developed for JBoss.
I have a somewhat fixed hardware configuration (servers have already been selected and they are meant to handle thousands of connections per second) and much more freedom for the software side.
I initially meant to use VRRP, either with dedicated L3 switches or via keepalived, and haproxy to provide access to Apache in a fail-over, loadbalanced configuration.
(Ref:
It is the first time i need to deploy JBoss so while doing homework i read that mod_jk essentially has a loadbalancer itself that would work between one instance of apache and multiple instances of JBoss.
My question is: If i configured both the http loadbalancing with HAProxy and the AJP/JBoss loadbalancing with mod_jk would i get any benefit? would it screw up the loadbalancing or even the functionality of the app server in any way?
Here is a diagram of the intended configuration:
We are still well into the design phase so this is why there are no more details. I can provide them if needed but i believe this to be a general question about the workings of different load-balancing mechanics when deploying JBoss + Apache.
I haven't taken databases into consideration as well, it is the next step we're gonna tackle.