[Babel-users] Questions

Mitar mmitar at gmail.com
Wed Mar 30 14:59:02 UTC 2011


Hi!

On Wed, Mar 30, 2011 at 1:00 PM, Juliusz Chroboczek <jch at pps.jussieu.fr> wrote:
> Section 3.4.3 -- Babel nodes compute the paths of lowest *cost*.  The
> important thing to understand is that the cost is derived from the
> rxcost and the txcost in a way that is implementation independent.

But it still requires that you can communicate directly in both
directions. Even if you would like to use for example only txcost?

> If you set cost=txcost, as in A.2.1, then you get shortest direct path, which
> is what you're asking for.

If I have a general directed graph and I know the costs of each edge
(which is txcost if we see it as a network) and I can send messages
only in the direction of the edges and I want to get the shortest
path/lowest cost path by Babel's approach, then even for such simple
graph as:

A ----c----> B

there is not luck for me, as B cannot tell to the A that it received
the message from it with the cost c and so that A has a link towards B
with the cost c.

So how B tells A this? If I understand only with IHUs? Which are send
only directly? So if there is a bad link backwards this could make
IHUs be badly transmitted and in the limit no IHU would get to A and A
would not know that there is maybe a very good link towards B? Even if
there is otherwise a good link somewhere else? Like:


A ----c----> B
|             |
A' <----d---- B'

So A and A' are connected with the ethernet cable, B and B' the same.
And you have one directional links between them. I am thinking about
very long HAM links which work only in one direction, using two
separate channels for each direction (so two routers on each side),
then Babel would not be able to do anything with those links?


Mitar



More information about the Babel-users mailing list