Is there a industry recognized tool to measure network latency between two VM's on the same host and same dvSwitch? Network latency is generally measured in milliseconds, but what is the expected latency for two VM's on a single host, assuming it's running at internal bus speeds?
This question is in regards to near real-time communications in HPC computing. I need to see microseconds to consider it near real-time, which is what I can achieve in my Infiniband HPC cluster. I need to prove or dis-prove the possibility of using VM's on a shared host for real-time simulations, without using an Infiniband fabric. I've already built and documented a 2432 core HPC cluster that's fully virtualized, hosting Windows 2008 R2 and CentOS Linux 5.7 x64 compute nodes, designed to overcome the inherent problem of irregular periodicity in high performance computing. (A VMworld 2012 submission)
Sorry for the shameless plug