I was wondering if ye could give me a little help, I'm doing a dissertation on virtualization and have to do some primary research, so my plan was to test the difference in performance between a physical system and a VM.
I had planned on loading up server 2008 on a real machine and a virtual 2008 on a HyperV 2008 server and than testing MS SQL performance FTP performance and Apache bench but my supervisor told me its not indicative of real world situation so does anyone know a better way for comparing performance.
Any help would be nice, Gav.
More about :benchmark virtual performance native performance
Ok what are you virtualizing and how are you going about the virtualization process will effect benchmarks. If your using W2K8 as a host OS then all your VM's will incur a performance penalty from the overhead produced by the host. If you use something like VM Ware ESX server then there is little if any overhead penalty. Something like Solaris 10 zones will incur zero host overhead penalty due to how they go about virtualization.
SQL can be virtualized fine using VM Ware ESX, even clustered. We've had big problems using W2K8 Hyper-V and SQL, we wouldn't recommend Hyper-V to anyone looking for an enterprise class service.