[ih] Some Questions over IPv4 Ownership

[Craig Partridge]

> > It's all store-and-forward, and the forwarding step uses whatever
> > external information the forwarder can glean.  A difference is the
> > routing mechanism.  (Typical) Internet routing gathers information in
> > advance because it has to forward its packets fast (and suffers when
> > reality doesn't match the information it has gathered), while mailers
> > have a nice mapping system built for them called DNS that they use on
> > demand.
> 
> I always thought email routing as end-to-end, if you take a message as
> a single data unit that message will be delivered (in theory) to
> whatever host has the MX record for the destination domain, instead IP
> packets are routed on a hop by hop basis with a chance of routing
> dynamically changing when the packet left the source and is in
> transit.

Go re-read RFC 974 -- you can route directly, you can route through
intermediate nodes.  All carefully designed to allow for failures and
failovers.  Also, your MX destination is not necessary the point of
delivery.  E.g. Email to bbn.com may be redirected internally to one
of several mailbox servers from which folks retrieve their email.
(Think of it as dynamic routing [MX] coupled with static routing
within an organization).

Thanks!

Craig