a question on policy gateways

[村山優子 <[e-mail]>]

We have been working on "routing by preference" for a while, 
and would like to ask this list a question on administrative domains
and policy gateways.

Routing by preference is concerend with how each one of the end systems
of a communication flow (i.e. a source and destination) can
enforce its preference of paths on a route --- to take 
the shortest paths in unfavourable parts of an internet.
We presume that those parts are administrative domains at the moment.
Each end system has its own favourites and unfavourites.

My question is:
Can the policy gateways in an administrative domain (AD) be 
monitored from the other ADs?
If so, does it mean that one can monitor some costs of an 
*intra-AD* link between policy gateways in an AD from another domain?
i.e. Suppose there are two ADs, W and S as follows:

     |           |
+----W3----------W4------+
|                        |
|                        |
|         AD W           |
|                        |
|                        |
+--- W1 -------- W2 -----+
     |           |
+----S1 -------- S2 -----+
|                        |
|                        |
|          AD S          |
|                        |
|                        |
+--- S3 ------ S4 -------+
     |          |


W1,...,W4 are policy gateways of AD W,
and S1,...,S4 are policy gateways of AD S.

W1 and S1 monitor each other because they are on the
same subnet (i.e. an Internet Exchange point),
so do W2 and S2.

How about W1 and S3/S4?
Should W1 be able to monitor the connectivity 
between S1 and S3/S4?
Alternatively, should we consider that W1 should get 
the connectivity information merely from S1 or any other
representative of AD S?

It looks like a network management problem as well,
but I thought it most appropriate to ask to this list.
If there is another list more appropriate, please let me know.

Yuko Murayama
WIDE Project