12-13-2013 09:02 AM
Customer question from John:
I’m working with a customer who is wanting to leverage dual hop FCOE in the following config:
Blades ---> VC Flex Fabric --> Nexus 5K --> Brocade DCX --> IBM Storage.
Per the Dual Hop FCOE Cookbook, a fully blessed and tested scenario is with an MDS at the FC layer and not a Brocade. According to the VC Product Folks, a Brocade can go there, however, the customer will not be able to leverage HP support beyond the Nexus 5K layer.
Ultimately, what we’re seeing is behavior that acts like NPIV is not on. Only one server / initiator is allowed to move FC traffic at a time. Turn that server off, then turn on another, and he works fine. However, once a new login comes online, the previous logins are squashed.
Any advice – anyone deployed against a Brocade, and if so, what was the magic sauce in your config?
Input from Dan:
I vaguely remember something about the Nexus 5K cannot be in NPIV mode and still do Dual Hop.
Reply from Julien:
NPV mode is necessary when you connect the Nexus to a non-Cisco fabric like in this example with a Brocade Fabric.
The NPV mode is a specific mode where all proprietary features are disabled to solve all possible interoperability issues, this mode uses NPIV to log into the Fabric. This is about the same mode we use with the FlexFabric and VC-FC modules for FC connectivity.
So from a technical standpoint, Virtual Connect Dual Hop FCoE with Nexus NPV mode (using NPIV logins) is a valid scenario and can be fully implemented. But as usual, the support from the storage vendor is another topic.
I received offline from John the customer’s Nexus configuration and their issue was due to a misconfiguration. The FCoE VLAN was not defined, vfc binding was wrong, etc.
Other comments or questions?