One of the biggest complaints I have with the Failover engine is the use of TCP Stack modification to suppress connectivity on the WAN link. A feature I would like to see would be using the built in "Clustering" capabilities of Server 2012 to create HA Clusters for Polling Engines. I believe this would greatly enhance the Failover Engine functionality and provide a better experience and transition when failing from Primary to Secondary.
If the above is too far out of the question, a version of FoE that would permit for adding a VIP (Virtual IP) that moves from server to server would be a useful addition to the FoE Software. Sure, it eats 3 IPs for a single 2 Node Active/Passive cluster, but the performance gain by simply "shutting the vip interface" on one server and "bringing the vip up" on the other, would be crucial. I know this would also pull into question how the system handles polling since all polling requests are outbound, my thoughts on this were to create rules that force all outbound polling to originate from the interface for the VIP, thus, no VIP no Polling.
Anyone else have any thoughts?
