I develop a TCP server application that listens to a specific port. One of my users is asking me if my application can run on an active-active Windows 2008 Enterprise cluster. Having only almost no experience in Windows Clustering, I googled and know that failover will work for Microsoft applications such as IIS and SQL Server, but I wonder if the same works for custom TCP application.
Firstly, I am aware that clients will connect to a virtual IP address instead of the physical IP addresses of the nodes, and if a specific node is down, Windows clustering will automatically redirect any connection request to the other node.
But how about the case in which only my TCP Server application hangs and stops responding (but Windows is still working), will failover still work? I might also intentionally want to reject the client connection if there is a critical application failure.
I would appreciate any pointers, thanks.