I've just read the new 3548 TR about Metrocluster and its architecture and I'm quite sure to have well understood that concept. I know fabric metrocluster for in the past I've set up several of them and, every time I was sure that there was a "dark fibre" on the ISL, a dedicated media or something that "come out" from some carrier multimodal line spilled out by a DWDM.
Reading at the recommedations on that TR I have theoretically a confirmation. See image below.
But, several times, it happens that customer declare to have some "FC/IP" line between distant sites, Cisco MDS 9513 and/or MDS 9222i used for other ISL using this "FC/IP" (I know this last one switch is in the MC matrix but I also know that the 4 switches must be dedicated to the MC!).
In this case I've often heard that this kind of line can be used for the ISL but at the same time that on those is not native FCP running but Ethernet at 10 Gbps that incapsulate FC frames... I also know that this kind of line can be used by concurrent synchronous replication mechanisms, such as EMC VPLEX Metro...that uses Ethernet that incapsulates FC or, if any, can use native FC.
So, this does not match with NetApp Fabric MC requirements.
Is there some detailed document that can explain this? We're not wan communication experts so we can loose ourselves in this kind of analysis.