What the heck is “FC Base-T”?
Posted in Storage Interconnects & RAID, Advisor - Tom by Tom TreadwayIt’s that time of year when all the Storage Gurus flock to beautiful Orlando, FL for Fall SNW. Personally, I think it’s a great show and it’s always well-attended by all the right storage folks.
To help spread the news to those that couldn’t attend I’m going to make a few quick posts on topics that I found interesting or puzzling.
I’ll start with a session on industry storage standards from David Black, Senior Technologist at EMC. David seems to have his fingers in a lot of pies, and always does a good job of summarizing new trends. He covered several dozen well-known topics in his short one hour presentation, but he brought up a new one that I had never heard of.
This new standard is “FC Base-T”. The proposal came out last August 2005, but I’d never heard of it - perhaps I live under a rock. There’s plenty that I don’t know. Apparently it’s FC protocol over a twisted pair, i.e., Ethernet cable such as CAT-5E. Interesting. That got me thinking about all the things that it “could be”.
For instance, is it just FC commands over Ethernet, sharing the wire with NAS, iSCSI and HTTP traffic? Then again, isn’t that just iSCSI? In fact, iSCSI and FC use the same command sets. Nope – that’s not it.
Or is the point that FC can now start using cheaper Ethernet switches? That would be cool. FC switches are still at least 2X the price-per-port of the more expensive Ethernet switches. Nope – wrong again.
The first reason that FC can’t run through Ethernet switches is that the lowest level protocol being used by FC Base-T is identical to the bits used on standard FC cables. The FC Base-T protocol is too different from the basic Ethernet protocol to be supported by standard Ethernet switches.
A second reason is that 1Gb FC does not run at the same clock rate as 1Gb Ethernet. FC starts at a raw 1GHz bit rate but drops to ~0.8Gb/s after overhead is removed. On the other hand, Ethernet starts at 1.25GHz and then drops to a real 1Gb/s after overhead is removed.
So I’m still a little fuzzy on the purpose of FC Base-T. It sounds like it’s mostly a cable-replacement technology.
Maybe some of the other readers could enlighten me as to the true purpose of FC Base-T. David Black brought it up, so I’m sure there’s meat somewhere on those bones.
Can I buy a clue?
TT
October 31st, 2006 at 5:38 pm
The Base-x notation in Ethernet is used to describe various physical layers, with T or TX used for twisted pair cable. So FC-Base-T would be a logical name for FC using a copper cable rather than an optical cable. Now FC was defined with an electrical variant all along, but that worked only at lower bit rates. Now that darn cheap 1G and soon 10G “Base-T” transceivers are are available, it makes sense to build FC switches and HBA’s that use these to lower the port cost. If you can’t beat ‘em, join them!
October 31st, 2006 at 7:20 pm
Ernst, thanks for the clarification. Do you know roughly how the price compares between FC-copper tranceivers and Base-T tranceivers?
November 1st, 2006 at 1:54 pm
Tom, such comparisons depend on many factors, like the type of optical link (multi or single mode, distance) and whether the transceiver is a modular plug-in (a GBIC or SPF) or built-in. To get an idea, a 3Com 3C996B-T Gigabit Server NIC retails for 130 euro, its optical brother the 3C996-SX Gigabit short range fiber Server NIC costs 540 euro, 410 euro extra, so in this case at least the optical parts are far more expensive than the rest of the product. For a more expensive product like a HBA or FC switch, the relative difference would be less, but still significant. Of course, the FC market is minute in comparison with Ehernet’s, so pricing will be less predictable.
November 1st, 2006 at 2:10 pm
Addendum: See http://www.t11.org/t11/stat.nsf/1158203694fa939f852566dc0049e810/0442390ee2d213068525706500739f06?OpenDocument for the FC-BaseT project definition, status, business case, etc.