Start a Conversation

Unsolved

This post is more than 5 years old

1730

July 27th, 2011 16:00

ESX4 with CX3 and CX4

Hello All,


I need to clarify a doubt regarding connecting a ESX host (or a group of ESX hosts in a cluster) to both CX3 and CX4. As per EMC recommendations, Failover mode 4 (ALUA) is suggested for CX4 running FLARE 28.704 or later. For CX3, it's Failover mode 1.


In such a scenario, where we have both CX3 and CX4 array to connect to a single host, what's the recommendation?


      Thanks,

      Rupak

61 Posts

July 27th, 2011 16:00

Failover modes (and the concomitant choice of multipathing policy) are on a per array / per datastore basis.

Its perfectly legitimate to set teh CX4 to FailoverMode=4 and RR multipathing on its datastores while leaving the CX3 at FailoverMode=1 and MRU or Fixed on its datastores - even to the same host.

29 Posts

July 28th, 2011 06:00

Thanks Matt..


I believe that the SATP needs to be VMW_SATP_ALUA_CX for ALUA configuration. Is that per LUN/device basis or system-wide basis?


    Rupak

July 28th, 2011 22:00

Since it is a host registration setting, it applies to all of the LUNs being presented from the array the host is associated with.  If you were to view each of the LUNs within vSphere Client:

1) those with registered paths set to failover mode=4 (ALUA) would all show as "Active" (but half are optimal and the other half are non-optimal)

2) those with registered paths set to failover mode=1, half would show as "Active" (current SP owner) and the other half as "Stand by" (peer SP)

And yes, as you point out, if the failover mode of 4 (ALUA) is properly being recognized by the ESX server, from the "Manage Paths" view it will list:

Storage Array Type: VMW_SATP_ALUA_CX

No Events found!

Top