I have the following HyperV setup I am configuring
Server with 2 teamed NICS connected to switch1 for LAN connectivity
Same server with 2 GB NICS set up on different LAN switch for iscsi and HP P2000 SAN connected with the switch.
iSCSI and MPIO has been installed and configured, but I noticed one of the MPIO path details showed source portal as 0.0.0.0/37320 and the other showe the correct IP address for the 2 NIC on the server. Is this correct? (attached screen shot)
What is the best algorithm for the load balance. Is Round Robin Subset the right one?
During my testing I noticed one NIC seemed to be used for the iscsi traffic., should that be the case with round robin?
I noticed was that there seemed to be a 20 second delay when testing a NIC failure by unplugging one of the iscsi NIC cables before the other would take over and the virtual servers would respond again. Is that normal behaviour? Can this delay be shortened?
Last question, is it best practice to set each port for iscsi switch to use jumbo frames? and is there noticeable benefit in performance.