> From: Toni Stoev <[email protected]>
> OK, locator names a place. The place is a node.
> The locator is formed as sequence of neighor IDs, each among the
> neighbors of the node with the preceding ID.
> This locator resembles a path but each component of it is a
> node-local ID, not a universal location name.
No, it's still a path - it's just that the names used for each component of
the path have a 'local' scope (i.e. they are not 'universal' names).
You are 'naming' locations with the paths to them - as I said earlier: "a
place is 'identified' (or perhaps 'functionally defined' would be a more
accurate way to put it) by a path which ends at that place".
You may find it desirable, for some strange reason (which I am utterly unable
to fathom), to call these things 'locators', but to the rest of us it's just
like calling a fish an 'eagle' - you can call a fish whatever bird-name you
want, but it still cannot fly.
A path is not a location, and _the properties of a path are not the same as
the properties of a location_ - an important observation which you seem to
be not taking into account?
Look, if you want to try and define a routing architecture in which you _do
not have_ a namespace which has the functionality/properties of what we all
think of as 'locators', and instead has only stub source routes to
destinations to 'identify' destinations, feel free. Such an architecture will
have a number of hard issues to solve.
For instance, the fact that the stub source routes which 'identify' locations
will have to change every time there's a connectivity change which affects any
element in those stub source routes. (I.e. local routing will not be able to
bypass a 'down' element - in other words, names will be very brittle, since
any local connectivity changes cannot be hidden inside the local scope, but
will have to be exported to all entities, all over the network, which are
communicating with any entities inside the local scope which include the down
element in their stub source routes.)
Not a good property. Perhaps you need to think this through a little more
before using more time here?
The use of stub source routes is not a wholly new idea, BTW - Dave Clark
wrote something about a similar concept (although his stub source routes
used global names) at around the time of an early IAB workshop on routing
(I think it was an IAB workshop on routing - not the 1998 IAB Routing
Workshop, but a much earlier one, in like 1987 or so - I can't find
anything about it online, but I remember Paul Tsuchiya and Van Jacobsen
both gave NAT presentations, Van on what he called L-NAT).
Noel
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
