[rbridge] it's time to summarize things
touch at ISI.EDU
Thu Dec 15 11:01:40 PST 2005
-----BEGIN PGP SIGNED MESSAGE-----
Gray, Eric wrote:
> We're getting closer, maybe, on this one. Please
> see below...
> --> -----Original Message-----
> --> From: Joe Touch [mailto:touch at ISI.EDU]
> --> Sent: Thursday, December 15, 2005 12:11 PM
> --> To: Gray, Eric
> --> Cc: Radia.Perlman at Sun.COM; 'Developing a hybrid router/bridge.'
> --> Subject: Re: [rbridge] it's time to summarize things
> --> Gray, Eric wrote:
> --> > Joe,
> --> >
> --> > I've addressed this issue before, just as you have
> --> > raised it before.
> --> >
> --> > Assuming that "external nodes" below refers to the
> --> > portions of the L2 broadcast domain that are topologically
> --> > outside of the RBridge campus, the RBridge campus actually
> --> > looks to external nodes as if it is a (possibly largish)
> --> > collection of L2 frame sources and sinks.
> --> For internal traffic, sure. For ingress/egress traffic,
> --> this would need to hijack ARP to work.
> Why? ARP, at least the ARP I am familiar with, allows for the
> conversion of an IP address into a LAN-local MAC address (much
> the same as DNS allows for conversion of a network name into an
> IP address).
> The RBridge campus forwards ARP - just as the local and remote
> LAN segments do - and the appropriate entity (an L2 source/sink
> such as a router or host) responds. Why would RBridges care
> about - or interfer with - ARP?
Because somehow the traffic from external nodes needs to arrive at
ingresses. Ingresses don't have (or at least don't use) external
addresses; the traffic from external nodes needs to arrive at ingresses
based on the external spanning trees, NOT based on being addressed to
In fact, there is NO external traffic that should ever be directed at an
ingress address - thus I believe that rbridge campuses do not
participate in ARP. That is a substantial difference from a router,
which DO participate in ARP.
Which means that the rbridge campus does NOT look like a collection of
sources and sinks to external devices w.r.t. ingress/egress traffic. The
only aspect in which rbridges DO look like conventional sources/sinks is
for internal traffic.
> --> To external nodes, the rbridge MUST act like a transit, not
> --> a source or sink.
> I think we're in agreement here. My statement was as to how the
> RBridge campus "appears" to the rest of the network. Yours is to
> how an RBridge campus "acts."
I'm confused about the difference. If all you mean by source/sink is
'traffic goes in that direction', then you're talking about
upstream/downstream channels, not sources or sinks. IMO, a source/sink
appears as a source/destination in a packet haeader, and (I now think we
agree) that never happens for ingress/egress traffic.
> Obviously the campus does not "act" like a source or sink. If it
> did, it would be a gibbering black-hole. It merely appears like
> a collection of sources or sinks to external nodes.
I'm not sure how that description is useful. Can you explain how?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (MingW32)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
-----END PGP SIGNATURE-----
More information about the rbridge