[rrg] Fundamental objections to a host-based scalableroutingsolution

[Tony Li]

Hi Eric,


|>> How can hosts make requests of the network today? The only direct 
|>> mechanism that I am aware of is QoS. 


IGMP and ARP immediately come to mind.  Various other IP options,
traceroute, ping, etc. ...

Take a look at the OAM discussion that's going on in CCAMP right now.  How
do we debug the network if we can't send requests from the host?


|The nature of my push-back to Tony is that I believe that the
|capabilities that Tony was ascribing to hosts aren't capabilities that
|actually are available to hosts at all. Rather, they are capabilities
|that are available to networks because of map-and-encaps (e.g., MPLS).


Let's be careful.  MPLS is an encapsulation, to be sure, but there's not a
lot of 'mapping' going on.  Sure, with an L3VPN or TE solution, you have a
FEC that forwards your packets down a particular LSP, but all of this is
based on pushed or configured routing information.  Not the same thing at
all.  As we've said before many times, the mapping is a very tricky part of
the equation.

Also, we should note that the reason that we get benefits from MPLS is that
it allows us to have a hybrid network architecture, where we can support
both connectionless (i.e. packet switching) and connection oriented (i.e.
circuit switching) styles.  If we go down the path of map-&-encap, we're
effectively deciding to run on top of a connection oriented infrastructure.
What then happens to the hybrid strength of the architecture?

Tony