[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RRG] Long term clean-slate only for the RRG?
Comments in line:
Noel Chiappa wrote:
My point was that, if my thinking is correct (and if it's not, can someone
please point out where I've gone wrong), it's not reasonable to
simultaneously say 'I want location/identity separation' and also say 'I
don't want to change existing host software' and 'I think a big mapping
database is not feasible'. If you want the first, one of the next two _has_
to give; the only question is _which_.
I phrase it in high-level terms, not in terms of any particular proposal, to
make clear that it seems to me to be a fundamental limitation. If I am
confused, can someone please straighten me out!
Sounds right to me. If there is going to be a split, someone has to
deal with the relationship between the parts. If the network is dealing
with that relationship, then that is a large mapping database. If the
host is dealing with it, then the host is changing (and may or may not
need a large mapping database.)
If not, then I think it would be interested to have people hum on i) do we
want location/identity separation, and, if 'yes', ii) do we a) change hosts,
or b) accept the need for a network-sized mapping database - making clear, at
the time we ask i), that it implies having to make a choice at ii).
It seems to me that separating out the way the communication system
looks at the identity of the communicating parties and the way it looks
at the location of those parties is a useful improvement in the system.
I am willing to accept host changes, if that is what it takes to get
there. (In fact, to really introduce the notion of identity there
probably have to be host changes. Identity ought to be constant across
the interfaces the host uses to talk as much as across the interfaces
the containing site uses to talk to the core.)
> I think the RRG has only focused on loc/id split and really hasn't gone
> into greater depth with other routing architectures.
Ah, I don't consider the location/identity split a 'routing architecture'.
It's an ancillary aspect of the overall system architecture, one which has at
best some influence on particular desired operational goals having to do with
routing (e.g. provider independence).
To me, routing architectures are about things like 'what entity choses the
path the packets flow along', and 'how are those paths computed', 'what data
does that computation need', 'how is that data distributed', etc.
I think that the question of whether the system has separate notions of
identity and location is an important architectural question. (I use
"system" because it is quite possible to construct approaches where the
identity is above transport, or is an opaque object used by transport,
rather than being below transport. Also, I use system because I do not
think it is either practical or effective to try to re-architect the IP
layer and its forwarding without looking at things around it, and
possibly making some changes there too.)
to unsubscribe send a message to email@example.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg