[manet] Re: non-manet islands as a large manet router

Hi David,

I am just seeing your message now after checking the list - I currently have a
problem where my IETF list emails are not being delivered to my inbox. It is
strange because a few messages are delivered while the vast majority are
not. The problem is also with all of my IETF lists and not just MANET - if
anyone knows how to diagnose and fix I would be very interested.

To one of your points below:

>> If manet border routers have some means of discovering each other, it's
>> certainly possible to do something similar on the internet, though it
>> would likely require (automatic) tunneling between border routers.
>> And whether it provides any benefit (particularly the "looks like one
>> large manet router" part) is a separate question.

With the AERO/OMNI link abstraction I spoke to in my talk, I call the things
you are calling "MANET border routers" as "Proxy/Servers". These elements
do represent each MANET as an aggregated network to the outside
Internetwork. And, as you say, it is based on automatic tunneling although
I prefer to call it simply "encapsulation". The encapsulation produces a
new layer in the architecture known as the "adaptation layer" which makes
MANET Internetworking possible.

Thank you - Fred

---

Hi all,


copy/extension of my mic note re. interworking manets:

there was some work (10+ years ago), targeting cases with small-ish
islands of non-manet networks (with or without internet egress)
connected to (or "embededd in") larger manet domains.

The approach taken there was for the manet border routers in the same
island to discover each other, relay messages of the manet routing
protocol between each other, exchange/synchronize their lists of manet
interfaces, and pretend to be one (possibly very large) manet router.
The manet border routers were also injecting manet host routes into
their non-manet routing protocol, which means no NAT or tunneling was
necessary.  Part of the rationale was that (a) the non-manet island was
zero-cost for the manet protocol, and (b) this implicitly handles
splitting and merging the manet domain (as in having such an island be
the only remaining connection) without any trouble or special approach.

(disclaimer: I was involved in coming up with the idea and suggested the
use of the IGP [OSPF there] to perform the discovery and synchronization
function and discussed it a bit, but dropped off quite early & wasn't
involved in actually implementing it and don't know how well it worked
in the end [it seems to have worked in general].)

If manet border routers have some means of discovering each other, it's
certainly possible to do something similar on the internet, though it
would likely require (automatic) tunneling between border routers.
And whether it provides any benefit (particularly the "looks like one
large manet router" part) is a separate question.

Cheers,


-David

_______________________________________________
manet mailing list -- [email protected]
To unsubscribe send an email to [email protected]