[Q - X-post wrong sub] Any diff btw network/TCP-IP stack btw Windows OS workstation and server versions?

Symptoms: delayed response forcing app level "complaints" - RSTs in TCP, or UDP missed heartbeats - on a network showing no packet loss in capture of transactions, just showing that the server app system is not responsive at times.

Can't figure out why we're recording two order of magnitude more problems, over the same network infrastructure, on a relatively high speed, high volume, small transactions environment, on the win 7 installed on a desktop server app, vs a server version of windows, running the same server app, installed as a VM, in a hypervisor based environment. Clients are the same. NIC characteristics, RAM, etc. very close in specs, on both environments.

Extra: what windows OS counters would you recommend monitoring, if to reveal $subj, or other problems in the network stack?