Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!elroy.jpl.nasa.gov!decwrl!sgi!vjs@rhyolite.wpd.sgi.com
From: vjs@rhyolite.wpd.sgi.com (Vernon Schryver)
Newsgroups: comp.dcom.lans
Subject: Re: FDDI question: bridges, routers, interoperability ...
Keywords: FDDI, bridges, routers
Message-ID: <104994@sgi.sgi.com>
Date: 18 May 91 22:48:52 GMT
References: <1779@ariadne.csi.forth.GR>
Sender: guest@sgi.sgi.com
Organization: Silicon Graphics, Inc., Mountain View, CA
Lines: 45

In article <1779@ariadne.csi.forth.GR>, nicolas@csi.forth.gr (Nicolas Chrissakis) writes:
> I have a host A connected to an ethernet segment. The ethernet segment is
> connected on an FDDI bridge and the bridge on an FDDI ring. I have a
> host B connected to an ethernet segment. The ethernet segment is connected
> on a FDDI ethernet router and the router on the ring.
> 
> My question is: Will A talk to B?
> The bridge will be Dec compatible translation bridge.

Ask DEC.  I bet it will work, provided you get the IP network numbers right.

> What can some one gain by having bridges mixed with routers?
> If routers are as fast as bridges why will some one use bridges?

It is impossible to build a "transparent FDDI to Ethernet" bridge.  For one
reason, the IP packet size on FDDI is 4352, which obviously cannot be
"bridged" to ethernet.  This means that any working FDDI<->ether "bridge"
is capable of at least IP fragmentation.  If you are bridging
ether->FDDI->ether, this would not matter, but there are many fast servers
that want to be on the dual ring itself, and with NFS/UDP/IP, cannot and
should not avoid using 4KByte UDP/IP packets.  Other reasons why DEC, for
one, sells what some call "translucent bridges" involve differences in how
data is encapsulated on ether and FDDI.  This second problem tangles up 
older versions, at least, of Apple network protocols.

The other side is that it takes fewer CPU cycles to bridge than to route a
packet.  It might take fewer cycles to IP fragment and then bridge
than to do full routing.

Some protocols are so stupid that bridging is the only reasonable solution.


> Do you have any experience with a router interfacing more than 2 ethernets
> and an FDDI? Did you notice or measure any performance degradation.
> I would like to have seperate measured filtering rates and forwarding rates
> for bridges and routers.

Demand real measurements of real benchmarks, not just promises.
There are sad rumors about current firmware or software for at least some
current FDDI<->multiple-ether bridges.  One vendor is making surprising
claims that the MTU on FDDI is 1500, contrary to RFC-1188, perhaps because
of performance problems.


Vernon Schryver,   vjs@sgi.com.