Hi Bill, |Have I grossly misunderstood how MPLS works? What's the story?
You have one model of how MPLS operates. However, another very common deployment of MPLS is for TE purposes. In this type of scenario, routing protocols are deployed natively and their information is used for path computation. Note that path computation need not be (and frequently isn't) simply just 'best effort'. A signaling protocol is then used to go hop-by-hop along the computed path and establish explicit state for each LSP. Again, this is significantly different than 'normal' tunneling (e.g., GRE) where only the endpoints hold tunnel state. Now, obviously MPLS does meet the 'encap' requirement, and for any tunnel to be effective, there is some mapping that directs traffic down the tunnel, so to be liberal, one could say that 'map-and-encap' is fulfilled. However, in a more strict interpretation, I would claim that map-and-encap architectures require a global mapping function that maps destination addresses to tunneled next hops and creates two namespaces: one inside the tunneled virtual topology and one underneath the tunnels. MPLS, when used in a TE scenario, is more general than that in that it does not create a separate namespace and it need not map soley on destination addresses. As always, this is a subtle semantic distinction, and the result really depends on your definition of 'map-and-encap'. While I love arguing semantics, I greatly prefer to simply select a common agreed on definition, make it semantically clear, and then move on. If further concepts need clarification, adding to the terminology is typically not a significant issue. Regards, Tony -- to unsubscribe send a message to [EMAIL PROTECTED] with the word 'unsubscribe' in a single line as the message text body. archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg
