Cisco MDS 9148 connection to HP C7000 Using Brocade Interconnect

[Marc Landry]

Good afternoon,

I am working on replacing our existing Brocade FC switches with MDS 9148. The MDS switches will connect to a HP C7000 chassis using the Brocade fiber channel interconnects (HP B-series 8/12c SAN Switch BladeSystem c-Class). 

Today the interconnects use ISL links to the existing Brocade's. After some reading I have come to the conclusion that the C7000 Interconnects should be placed in Access Gateway mode. This seems like the simplest approach to me.

On the Cisco side I just need to enable the NPIV feature correct?

# show feature | i npiv
npiv                  1         disabled

Will all the Cisco features play nice with Brocade in this mode? Things like smart-zoning? Are there any limitations I should be aware of?

Anybody running a similar infrastructure that cares to share their experiences?

Thanks,

[dynamoxxx]

I have my C7000 enclosures with Brocade 8/12 connected to MDS 9513. Like you said, enable NPIV on MDS and enabled AG on Brocade (that will wipe Brocade switch clean). This is the easiest way to connect those Brocade switches.

[Marc Landry]

thanks for answering.

Once AG enabled you simply see the blade WWN's in the flogi db?

Nothing to change in the C7000 on board administrator?

[dynamoxxx]

exactly, once you enable AG (ag --modeenable, switch will reboot after this command) host WWN will show up in "sh flogi database"

Do you plan on maintaining storage connectivity during the migration, working on one fabric at a time ?

[Marc Landry]

Yes that is the proposed approach. I have a mix of VMware hosts and bare metal (mostly Windows) boxes. I will take down one fabric at a time. Here is the high level order I was planing on following:

• Turn OFF "Big" Brocade #1
• Put C7000 Brocade bay 3 in AG mode
• Move Storage Frame & C7000 cables to Cisco MDS first (Test this)
• Migrate bare metal server connections

Repeat process for the other fabric a couple of days later so we can allow for a soak period.

Do you think there is an advantage in forcing all the ESXi host to use the other HBA during the maintenance window, or just let VMware round robin/Last used do its thing?

[dynamoxxx]

plan looks good to me.

I would let ESX multi-pathing do its thing on its own, it's pretty robust.  How about your physical hosts, are they using MPIO or other vendor provided multipathing ?

[Marc Landry]

MPIO for physical. Any gotcha with this? Anything to change or prepare?