Folks, can anyone here please share the best case scenario of using Unicast and Multicast on Windows NLB implementation on VMware ESXi server ?
As far as I understand, the Unicast NLB is supposed to be used by web server who needs to be on both nodes to be Active / Active, and it doesn’t matter when the reply comes from the different server since the content is the same and no user session is persisted.
To be used with Non Persistent Windows Terminal Server, when the user connect to the Cluster Virtual IP, it will then randomly selects the available node or Server A in this case, once they disconnect and then reconnect again, different server might serve them (Server B).
Multicast or IGMP Multicast
To be used with persistent connection server while still maintaining High Availability, for example Exchange Server 2007 Client Access Server. Because when the user connects to the CAS server through Active Sync protocol, in this case Server A, it needs to maintain the session and the connection until the user is logged off of the session is closed gracefully, the user will always be connected to the Server A no matter what. In the case of Server A outage, the other server in the node in this case Server B can quickly take over the service until it is drain-stopped and Server A back to the operational state.
Is that analogy correct or needs some clarification?