We have a Server2008 host, and three virtual Server2008 machines within it.
The host has a dedicated NIC, the three virtual servers share one NIC, and
all is well, sort of.
When the configuration is alone, (not attached to our corporate network,
essentially on their own wire), all three virtual servers network well, ping
is quick, no packet loss, our service APIs between each/all of the three
servers behave fine.
After we attach our Server2008 host to our corporate LAN, bringing it into
contact with our Cisco switch and the rest of the network, things change.
The host OS is fine to the corporaate network, and the vrtual OSs seem fine
to the corporate network. But, the the three virtual servers communications
with each other get very spotty. In particular, if traffic among the three
virtual servers goes idle for a short period, initial pings get dropped, and
our service APIs timeout. As long as we keep steady traffic between the
virtual servers, (via pings beween each/all virtual servers), things remain
fine. But after a short period of quiescence, trouble again. (A theory is
that the ARP in the virtual switch for the three virtual servers is getting
blasted somehow, overridden incorrectly from the corporate switch. I don't
know, I'm a C++ and SQL Server guy.)
Any thoughts or comments will be very much appreciated. I hope this is
somewhat coherrent, and thank you in advance for reading this.