I'm having a debate with other network engineers about the benefits of using load balancers at the Data Center. The client requires N+1 redundancy at the datacenter, which means we'll need to deploy four WAVE 7471 devices at the main datacenter, and four more at the DR site.
WCCP V2 seems to handle load balancing among WAVEs fairly well, when deployed in cluster mode. Except that in the event of a WAVE failure in the cluster, it could take up to 49+ seconds for WCCP V2 to restore TCP session flow, for those TCP sessions that were being handled by the failing WAVE device. By the time 49+ seconds elapses, some TCP flows would have exceeded their retransmission timers.
Deploying load balancers seem to help with that issue, as they can protect TCP flows with faster session redirection.
Thoughts?
TIA
Angel