I am about to purchase servers for a virtualisation project and still have one area of uncertainty.
The vmdk files will be stored on the iSCSI server along with the data. I was originally thinking of having the ESXi servers with no disks and the hypervisor installed on an SD card. I was told by a supplier yesterday that the most usual approach is to still have the vmdk files on the SAN but to have 2 disks in RAID1 on each server and to have the hypervisor installed on this array. The reason being that it moves the processing from the SAN server to the ESXi server.
Please can tell me if this is correct and explain why?