The Routing Over Large Clouds Mailing List Archive by date

[Keith McCloghrie <kzm@cisco.com>]

> This, in a nutshell, is one of the big NHRP issues: the client is in a 
> position to know how much it wants to pay for the connection but the network 
> is the only place that knows how much it will cost. With NHRP as the
> signaling protocol, there is no way to set up a mutually beneficial deal.
> Even worse than this is a situation with an intervening "low-cost"
> proxy-client
> which probably does not communicate any form of cost information with the real
> client (who is the one with the open wallet).
 
If the "real client" is an NBMA-attached host, then presumably the only
way it can get any "cost" information is to implement the NBMA's routing
protocol, e.g., PNNI, but to have hosts implement routing protocols is
contrary to our networking paradigm.

If the "real client" is, say, an Ethernet-attached host and there's a
router between the Ethernet and the NBMA network, then it's not clear
to me how the router can communicate "cost" information to the host,
but whatever method is used would seem to be independent of NHRP.

If there's an MPOA "low-cost" Edge-device between the Ethernet and the
NBMA network, then that Edge-device is part of a distributed router,
and that distributed router is responsible for communicating the same
"cost" information (whatever it might happen to be) to the host.  So, I
don't see the significance of the cost of the intervening device.

Keith.