Exchanges that matter...

It is actually quite hard in practice to push a FDDI switch
into trouble, although theoretically it should be simple, modulo
the interesting effects that token-withholding can produce.

Hm. Not sure what you mean. FDDI can certainly be over-utilized, that's
not hard. On the other hard, FDDI doesn't fall down and die like
Ethernet does when congested.

In practice, the limiting factor will tend to be the per-port
bandwidth rather than the aggregate bandwidth. One would have


At the Atlanta-NAP we offer full duplex FDDI, why not try to get MFS to do
it? Cisco now has a full duplex FDDI card, so you can do 200 Mbs into the

Nathan Stratton CEO, NetRail, Inc. Tracking the future today!

No, you can do 100Mbps _into_ the peering point, but you can also pull
100Mbps out of it at the same time.

But I agree, it would be nice if MFS enabled full duplex FDDI on their
gigaswitches. Unfortunately, I think only 75xx series routers can use
the full duplex FDDI card, so people will have to upgrade routers. I
could be wrong about this, though.

However, this does not address the problem of the peering point itself
getting overloaded (as has happened at MAE-east). MFS seems to have
moved all of the big traffic connections to the first gigaswitch,
which has actually made things much better. But this is only a
temporary solution at best.