Hello -
I recently setup an environment utilizing 8 IO Analyzers each accessing either an iSCSI LUN or NFS Share (but not both at the same time). The secondary virtual disk was set at 30GB.
For the iSCSI tests, we ran a 50/50 0% random workload and our total IOPS reached 4086.11.
When we created the second disk on an NFS datastore and ran the same test as above, our total IOPS only reached 583.57 for the same time period (2 hours). Additionally, latency doubled.
I checked IOMeter on the guests, and it appeared they were pushing fewer IOPS as well.
Any ideas as to why we couldn't push as many IOPS using NFS? I would think the amount of IOPS would be the same across tests, regardless of the backend?
Thanks in advance.