HP c-Class BladeSystem

[Jilani-Malik]

I'm configuring 8+8 full-height BL620c G7 blade servers in two stacked HP BLc7000 enclosures. For the first c7000 enclosure, I've provisioned two HP 6120G/XG Ethernet switches and two Cisco MDS 8Gb FC switches. I just need to confirm if I can use the unused ports of the ethernet and SAN switches in the first c7000 enclosure to support blade servers installed in the second adjacent c7000 enclosure. If so, do I have to use pass thru switches to implement such an arrangement?

[Casper42]

Stacking as the term is used for Virtual Connect does not immediately apply to non VC modules.

So the question really is what is the best way to connect these 2 chassis to the network and SAN.

First for the SAN. If you have the MDS Switches in NPIV mode then as far as I am aware you cannot nest them. I wouldnt reccomend it anyway because FC is a place where you really dont want to introduce latency if you can avoid it. If the switches are in normal switch mode, you can daisy chain them but you will lose some of the uplinks on the first chassis in order to connect to the 2nd chassis. My reccomendation would be to home run all your FC cables back to whatever your central FC Switch is, from each chassis. Then they can be in either NPIV or normal Switch mode.

For Ethernet, you can daisy chain the switches together, but you will need to keep in mind things like spanning tree and whether you are really Active/Active or not. I would reccomend home running these with 10GB uplinks, but if such uplinks are a premium, I would run the Bay 1 uplinks through chassis 1 and then daisy chain to chassis 2 (DAC Cables are perfect here) and then do the opposite for Bay 2. Uplinks in to Chassis2:bay2 and then daisy chani back to Chassis1:Bay2. This will give you optimal redundancy, assuming you have Spanning tree and server side Teaming/Bonding/Failover configured correctly.

Ultimately just keep in mind that whatever you do, because of the switches you have chosen, there are no major rules changes between blades and rackmount servers with top of rack switches. So however your Network and SAN teams decide works best can be implemented with blade chassis based switches as well. There is no magic with respect to I/O just because they are blades.