UCS connecting to SAN switches

[tbone-111]

I've configued UCS direct connecting into the SAN before, but not SAN switches.  My understanding of the main differences between the two is this:

1) With SAN switches, you configure vSANs.  To do this go to SAN tab -> SAN Cloud and configure your vSAN ID's here.  Then in your vHBA template, configure it to use the appropriate vSAN ID you configured under SAN cloud.  Anything else to do here?

2) Whereas with direct attached storage you connect FI-A to SPA and SPB and FI-B to SPA and SPB, with san switches FI-A goes to 'switch-1' and FI-B goes to 'switch-2'.  Is this correct?  If so, anyone know why it has to be this way?

[Trevor Roberts Jr]

Regarding #1, nothing else needs to be done except configure your FC uplinks on the FIs.

Regarding #2, I never learned the official reason why it's supposed to be that way between FI and SAN switches. However, traditional SAN best practices for redundant fibre fabrics is to have your A and B (or odd and even) switches connected on completely separate paths. It's only at the end points that you mix and match(ie your array SP ports, and your service profile has vHBAs from A side and B side.

If you have MDS switches, you can take advantage of FC Portchannels as well when you follow this design.

Sent from Cisco Technical Support iPhone App

[tbone-111]

Great, thank you!

One last thing, when configuring my boot policy, how would you configure the SAN target primary and secondary if you have 4 HBAs per fabric interconnect?  Can you create multiple SAN secondary ports per vHBA?  Or is it only supported to have 2 per Fabric Interconnect?

[Trevor Roberts Jr · Accepted solution]

No problem, tbone-111

As far as I know, only two vHBAs are supported at this time.  It's best to use 1 from the A-side and 1 from the B-side.

The vHBAs that are used depend on which wwpns you zoned to your boot luns.

For the example below, assume the following convetions:

1.) my vHBAs are named, vHBA_A1, vHBA_B1, vHBA_A2, vHBA_B2 (A for A-side fabric, and B for B-side)

2.) vHBA_A1 is the first HBA on my A side, and vHBA_B1 is the first HBA on B side

AND

3.) These HBAs' wwpns are zoned properly to my boot LUNs' Storage Processor (SP) ports on the FC Switch

AND

4.) These HBAs' wwpns are LUN masked properly on my array

Those are logical ANDs, by the way, they must ALL be true to prevent headaches later on.

5.) Disk Array SP ports A0 and B1 are connected to my A Fabric FC switch.  B0 and A1 are connected to my B Fabric.

6.) This example boot policy will have the A Fabric be the primary boot path

-FYI, best practice to have at least 2 SAN Boot policies (1 for A-side and 1 for B-side, and evenly distribute among your Service Profiles).

In my boot policy, I check the enforce vHBA naming option and specify my vHBAs names (sorry I don't have screenshots available.  just started with a new company and don't have lab access yet).

My SAN Primary setting might be: Primary Target: A0's wwpn and Secondary Target: B1's wwpn

My SAN Secondary setting might be: Primary Target: B0's wwpn and Secondary Target: A1's wwpn

It would be a good idea to confer with your Storage Admins while you are setting up the first couple of servers to boot on each Fabric to help with any troubleshooting.

If zoning and LUN masking are configured correctly, you should see WWPNs listed during the UCS Server Boot sequence at the point where the FC Boot driver is being loaded.

Of course, get Cisco TAC on the line if you run into any obstacles that you cannot get past.

I hope that helps,

Trevor

======================================================================================

If my answers have been helpful in any way, please rate accordingly.  Thanks!

Message was edited by: Trevor Roberts Jr

[tbone-111]

In talking to my SAN engineers, apparently they don't have anything like a vSAN ID I can enter into the UCS chassis.
  After I've created my service profile, I get an error on each server stating that it can't see the WWPN's of the array.  What other steps need to be done?  I'm guessing that unlike direct attached storage, the SAN engineers will need to zone the WWPN of each blade server as well before it will see the array, is this correct?

And back to vSANs, since they say they aren't using anything like a vSAN ID(they are using HP switches, I believe), what should I put in here on my vHBA templates?  Right now I just have it set at "default".  Do I have to add this "default" to each Fabric Interconnect as well, perhaps? 

Thanks in advance!!