I'm in the process of setting up a Xenserver pool that uses a shared iSCSI drive to store virtual machine disk images(An iSCSI drive is a raw partition or block device that's exported over a network, kind of like a NFS share).
I'm curious as to what would be considered best practice for benchmarking this iSCSI disk. I have created a Centos 6.2 virtual machine with Phoronix-Test-Suite 3.6 installed. I have run several disk benchmarks but several of the results are unrealistic and claim to be faster than the network could possibly handle. I presume this is due to local caching of the disk.