I am currently working with a customer that has 3xHP BL20P blades a 2xCisco Catalyst 4948 switches. Each blade has 4 physical switchports of which 2 are plugged into 1-Cisco Catalyst 4948 and the 1 port is plugged into the other Cisco Catalyst 4948. The 2 switchports are plugged into a switch that is accessible by the production network and the 1 switchport that is plugged into the other switch is used for an isolation environment for iSCSI traffic only. The SAN unit is the LeftHand SANIQ v7 loaded on HP Proliant 320S x3.
The problem that the client is experiencing is severe instability in the environment to the point that upper management is beginning to loose faith in the VMware environment and is considering going back to physical servers. We all know that it is a waste of time and effort to back and I am fight for the customer to revive their faith in VMware. I suspect that the problem is self inflicted due to the fact that they have some VM guests that also have iSCSI volumes being mounted via Microsoft iSCSI initiator to the Windows 2000/2003 VM guests using the same physical switchport as the ESX hosts. Can I get the communities opinion regarding this configuration? I want to get the experts opinion to drive efforts to use the unused switchport on their BL20P blades to dedicate that to iSCSI traffic via Microsoft iSCSI initiator for their Windows 2000/2003 VM guests. Thoughts?
What instability are you having? slow performance? disconnects?
How many VMs?
Have you checked simple things like duplex/speed on the LeftHand?
Are you using single nics on the Lefthand or mutiple nics with ALB/802.3ad?