Been thinking about running IP on a DCX. What are the pros and cons of running IP on a FC switch?
Haven't been able to find much documentation on how to do it, we have storage only on the switch now and would like to isolate IP traffic to just those HBAs that need it, would soft or hard zoning work here, or would I need to set up a VF environment to keep IP broadcasts away from ports that don't need it?
In the past if found only certain FC cards (2G or lower) being capable to do this drivers wise.
I'm not sure but I guess IPFC is not financially viable anymore with +10G Ethernet.
we don't have enough 10G ports to cover the effort, but do have enough ports on the DCX is the reason we're looking at it.
Well you need to find a FC IP device driver for your card, perhaps there are OSS frameworks.
Install and configure them then config IP
On the switch you need to zone the two HBA together, nothing else should be needed, although I could not find evidence that current FOS and Gear support http://www.ietf.org/rfc/rfc2625.txt nowadays.
Sorry for not having anymore detail, but I've never tried to attempt IPFC.
Perhaps someone on the forum with IPFC experience can help you out in more detail
In general, I'd advise against using IPoFC. It has never been embraced in the storage community. For that reason, HBA device driver support for it is "crusty" at best. Or, said differently, not many frames are flowing in SAN fabrics using IPoFC, so that part of the code streams aren't exercised much, if at all.
The anticipated equipment cost savings may well be eliminated by the labor cost of figuring it out initially, then getting support whenever something doesn't work, and the stress of being "one of the few" doing it, particularly in a production environment.
Just my 2 cents worth. And yes, I'll glady receive change back :-)
I appreciate all the responses