Paul Koning wrote:
FDDI and 100BaseT are the same speed given equally
competent
implementations.
With 100BaseT using an MTU of 1500, and FDDI using an MTU of 4478? I
don't believe there's a standard for jumbo frames on 100Base-T.
Sridhar> That's not even talking about
fault-tolerance.
Ethernet and FDDI both offer fault tolerance. They do it in different
ways, but they both can do it.
Having spent several years in the FDDI ANSI committee, I know all the
"why FDDI is better" FUD. It doesn't hold water. And it certainly
doesn't justify the absolutely mindboggling complexity of FDDI
compared to Ethernet.
Take a bunch of SAS CDDI cards, plug Cat 5 into them, plug other end
into concentrator. Voila. Done. Where's Need to run it a couple miles
down the road? No problem. Not everything has to be backbone dual
counter-rotating rings with trees.
That, and now that the hardware is out of fashion, it's a lot cheaper.
I have some GigE and some FDDI running over fiber between buildings here
at work, and the GigE's redundant failover capacity often takes special
hardware ($$), and often doesn't work as well.
> You can
run IP over it, of course. And then you could run iSCSI,
> if you don't mind the slow performance. But FDDI was obsolete
> long before iSCSI came out. So an FDDI to SCSI converter would
> have to be a proprietary hack.
Sridhar> Obselete? It's still very much used in places where fault
Sridhar> tolerance is important.
You're thinking about the U.S. Navy? I heard from them a few years
ago when they realized that FDDI was dead, dead, dead, and they had to
look for alternatives, because new FDDI hardware could no longer be
obtained anywhere.
It's all over IBM's internal network NOC's. I seem to remember one of
the larger ISP's fairly near here still using it in their NAP. I could
probably find out who, if necessary.
That, and this is classiccmp. Old != obselete.
Peace... Sridhar