How to check the UCS SAN connectivity is redundant?

[SrikanthV]

How to make sure the SAN connectivity is redundant. How does SAN failover happen in UCS environment?

[Walter Dey · Accepted solution]

I assume, your design is redundant

eg. 

- each FI connected with native FC to a MDS or N5k (no dual homing)

- therefore physical dual FC fabric

- separate VSAN's (=/ 1) for Fabric A (VSAN-a) resp. B (VSAN-b)

- Service profile with 2 hba's, connecting to fabric A (VSAN-a) resp. fabric B (VSAN-b)

- on the Host OS install the proper multipathing software

- check that your multipath software see's a LUN over 2 paths.

- check Failover, by removing cable connecting FC to MDS ? or disabling vhba for a fabric.

[Kirk J]

Greetings.

Unlike the Ethernet vNICs, vHBAs do not have UCSM fabric failover capabilities by design.

As Walter's check list explains, the dual Isolated SAN fabrics is a general industry best practice.

It is incumbent on your OS to be storage multipathing aware.

Most OS multipathing/NMP settings provide for redundancy fail-over as well as load balancing.

For ESXi, for example, you can individually disable specific paths, pull cables/disable storage uplinks, disable vHBAs (already mentioned by Walter).

UCSM also has a Fabric evacuation option, that will drop all Ethernet and storage UCSM connectivity for a specific FI, which is a good way to have a large scale multipathing fail-over test.

Thanks,

Kirk...