SAN Zoning configuration on Cisco UCS
Posted by craig
- on March 13th, 2010 in Data Center, Hardware, Server, Storage, Tips | 10 Comments »

Recently I had deployed quite a number of EMC SAN with MDS Switches and Cisco UCS, I found tha FC SAN zoning might be a key consideration we may need to take a look for every deployment. This post will more focus on the MDS and UCS 6120 FC Uplink.
Cisco UCS interconnect fabric switches are utilizing FC uplink to allow the Cisco Unified Computing System to get access to the SAN fabric environment.
Here is the architecture diagram you can refer to. In order to make this work, 1st requirement will be the NPIV.
What is NPIV? ( N_Port ID Virtualization or NPIV is a Fibre Channel facility allowing multiple N_Port IDs to share a single physical N_Port)
For more details information, you can refer to the URL HERE
Once the NPIV had been enabled on the SAN Switch, you just need to enable the FC uplink from the Interconnect Fabric by the UCS Manager or CLI mode. Now you should able to see the WWN login from the command prompt, device manager or fabric manager. Any 1 of the method will work. To simplify the setup for the UCS, you can consider port base zoning or WWN zoning. I had done both setup before with VSAN in place, it looks similar to me except from security stand point. If we are only connecting host and storage via the SAN, the port base zoning with the Storage host group features should able to manage the host group to access the correct LUN. If there are additional devices such as tape library, this may create another layer of confusion. At the end, I configured all the zoning base on WWN id, which are still the best configuration in most environment today to prevent any device, host or target conflict in the fabric environment. Not to say the port base zoning will not work, it just the WWN base is more secure and reliable.
I had done the test with the port base zoning and end up, the backup device will be visible even from my ESX server which run on top of Cisco UCS. This is not cool


10 Responses
Craig,
Thanks very useful. I’ve been struggling with this for some time. How do you see the the WWN login: either via cli, device manger or fabric manager. Where exactly you do this? And also how do you mask it in Clarion (or any storage).
in MDS you can try this command, show flogi database, or show flogi fcns database
in Clariion, I created the host group and add the necessary host to access to the specify LUN. All the host WWN will be required to be registered in Navisphere from EMC.
according to Cisco’s recommended configurations, WWPN is the preferred method. WWNN mapping can cause issues. Review here:
http://www.cisco.com/en/US/docs/unified_computing/ucs/ts/guide/TS_SANBootAndConnect.html
From the URL you posted, it does mean that. In all deployment we done, we did not include WWPN, so far everything is running fine. Let me verify with my counterpart from Cisco again about this and come back to you. Thanks for sharing.
Sorry as I had confused with the configuration, you are right, for the SAN boot, we did include the WWPN for the storage target.
Thanks for this article. This is the only one I could find online about the Cisco MDS and Cisco UCS uplink configuration for the Fibre Channel Network.
I have a quick question on proper configuration when zoning though. When connecting the two UCS devices to the two Cisco MDS (9100) switches should I create two separate VSANs such as VSAN10 and 11 for each individual UCS ? Or can they both go into the same VSAN ?
When you mentioned 2 UCS, let’s assume you will have total 4 UCS Fabric Interconnect 6120 in this case. If you would like to access to the same VSAN for both UCS sets, you may consider to configure same VSAN for both UCS.
This may happen when all the necessary Storage Devices require to presented to 2 UCS sets. In another situation, you may consider to build a new VSAN with VSAN routing on MDS for isolation purpose. There is no right or wrong, choose the best fit for your case.
Thank Craig, that makes sense. Thanks for the reply
Thanks for this. We just got UCS plopped in our env and currently not using any NPIV. Any one else experience any downstream affects? Curious.
How many uplink did you connected from each Fabric Interconnect?