This article is from the Token Ring FAQ, by James Messer James@NetworkUptime.com with numerous contributions by others.
Source route bridging is a method to connect two similar network
segments to each other at the datalink layer. It is done in a
"distributed way" where end-stations participate in the bridging
algorithm, thus the name _source_ routing. (as compared to
transparent bridging, refer to 5.9]).
In a source-route bridging environment a source end-station will
sent out a "route explorer" frame (broadcast) to find out the route
to the destination end-station. Source route bridges will forward
these frames to all segments/ports. The source route bridge will add
route information (the segment the packet came from) to the frame
prior to forwarding it. This route information is called the Routing
Information Field (RIF).
Eventually, the route explorer frame reaches the destination
end-station INCLUDING THE COMPLETE ROUTE (via the RIF) the packet
took. The destination end-station then uses this RIF to reply to the
source end-station directly (i.e. no broadcast). Please note that
the reply traverses all bridges in reverse order of the route
explorer frame and INCLUDES THE RIF. When the reply reaches the
source end-station, the complete network route is known by both the
source and destination end-stations. Subsequent packets will use
this route information (i.e. no broadcast).
It is possible that a network has multiple routes to a destination
end-station. In this scenario, the source end-station will receive
more than one reply to the route explorer broadcast. In most cases,
the source end-station uses the route that was received first.
In a source-route bridging environment, the end-stations discover
and store information about the network topology. In a transparent
bridging environment, the (transparent) bridge discovers and stores