mod_caucho vs. resin load-balancer
Hi,
For reasons of our business we need to use an Apache webserver front-end to our Resin 4 cluster.
We are using mod_caucho to route Resin traffic to our cluster. These as oppose to proxying the traffic to a resin load-balancer process. We figured this to be more efficient as fewer IO will be produced.
We are seen some estrange behaviour in the cluster where one of the 3 clustered resin servers gets most of the load, while the others are mostly unused. Mod_caucho does not seem to be distributing the load evenly between the cluster.
Is mod_caucho as good as the Resin load_balancer on distributing the load across the cluster? What is the best strategy when an Apache webserver must be part of the pipeline.
Jose Alberto
|