Hi Steve,
thanks for your reply. I think I can clarify this for you.
You're talking about the 1/2/4/8 Gbps 48-Port FC Module, part number DS-X9248-96K9.
Look closely at the part number, it says 96K9. The 96 refers to the 96Gbps uplink to the backend this linecard has.
12Gbps * 8 portgroups = 96 Gbps. The .8 is there, so in reality you even have a little bit more bandwidth like you calculated above. It's just rounded off here for ease of viewing.
So why do our marketing documents specify 192Gbps? You'll see that this figure is closely followed by 2 very important words: Full Duplex. You have to interpret this as 96Gbps up and 96 Gbps down, good for 192Gbps total. A 4Gbps port will be able to send and receive at the full 4Gbps in each direction.
I agree that this is confusing. Unfortunately this is not going to change, simply because this is the way everyone does it in their marketing documents.
The way to use this linecard is hidden in the notion that it's 12.8Gbps bandwidth is shared, and hence you can already start connecting 8Gbps devices to your director switches by carefully designed your SAN so 8Gbps hosts share a port group with less demanding nodes.
A much higher bandwidth 8Gbps linecard and accompanying Supervisor linecard are on the horizon.
I hope this clears things up for you. You might want to also take a look at our next-gen Consolidated I/O 10Gb FCoE switches, the Nexus family. The Nexus 5596UP offers a non-blocking architecture unlike the one in the MDS 9500 and will also run plain FC ports. I'd recommend you to check with you Cisco Account team to see what exactly the differences/tradeoffs are and if this is a viable alternative for you.
