Re: portfast

From: Jim Warner (warner@cats.UCSC.EDU)
Date: Sat Jul 07 2001 - 18:23:18 EDT


| i don't see the point of running STP on any port that's attached to a single
| host. i only run it for the ports connected to my routers, hubs, CSS, PIX,
| trunks to other switches..
|
| am i missing something or is this kinda the dog chasing his tail?

First, you need to know the effect of a topological loop if you
don't have spanning tree. Our experience is that the switch
completely locks up, using all its BW to feed the loop. If
you've got a big L2 network, you'll be out there disconnecting
jacks one at a time looking for the loop. "Portfast" and friends
from other vendors will permit a loop to form as a transient and
then clip it off.

At Universities, the switch ports are in the hands of our students.
And our users will eventually hook everything together in all
possible combinations -- as long as the connectors fit. And
when we finally track down the student and ask them why they did
it, we'll get some lame answer like "I didn't want to let the
connector fall on the floor while I wasn't using one end so I
plugged it into the spare wall jack."

Your intention that any particular port only goes to a "single host"
can be blunted by a visit to the Microwarehouse catalog. Any user
can show up with a repeater and go into the connectivity business.

Perhaps STP is like President Bush's hopes for his missle shield.
It's not going to some some with real malice. But it's enough
to deflect the clueless. [I am not making any claim about the
relative merits of ABMs and STP -- except that I'm sure STP costs
less.]

-jim



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:44 EDT