ietf
[Top] [All Lists]

Re: [mpls] Last Call: <draft-ietf-mpls-in-udp-04.txt> (Encapsulating MPLS in UDP) to Proposed Standard

2014-01-23 08:49:54
    > From: "Eggert, Lars" <lars(_at_)netapp(_dot_)com>

    >>> Depends on the reach of that encapsulation. .. If it traversed such
    >>> middleboxes by default, as UDP does, it has the same issues.

    >> Exactly. The problem is not specific to UDP.

    > you conveniently cut the part of my original message where I tried to
    > explain that the difference is in *reach* of the traffic.

Say what? Does not "Depends on the reach of that encapsulation. .. If it
traversed such middleboxes by default" (which I did include) make exactly
that point?

But that's not important. Moving on...


    > encapsulators are not forwarding nodes from the viewpoint of the
    > network they tunnel over. They are traffic sources.

You might just as well say that 'from the viewpoint of the rest of the
Internet, a NAT box is a traffic source'. I mean, as far as the rest of the
Internet can tell, it just sits there, and emits packets, right?

After all, i) the packets it emits are quite different from the packets it
receives, and ii) (to make your pet point) the original packets (which
usually have RFC-1918 source addresses) would not make it far through the
Internet without the header changes the NAT box makes. 

But I don't hear you clamouring to have NAT boxes detect, and respond to,
down-stream congestion.


Look at it from the point of view of the box. Something(s) is sending it
packets, which it is supposed to forward. Basically, it can do one of three
(really, two) things with the packet:

- Forward it
- Drop it
- Queue it (but this eventually has to turn into 1 or 2, above)

How is that any different between i) a plain old router, and ii) one of these
boxes that you insist is a "traffic *source[]*, not [a] *forwarder[]*"?


You want to separate out a very specific class of intermediate forwarding
nodes, selected on a fairly arbitrary basis (the particular kind of header
they emit), and impose requirements on them that are not imposed on any other
intermediate forwarding nodes - and change the Internet's basic schemea of
congestion control in the process.

Why restrict that change to this particular kind of intermediate forwarding
device?

        Noel

<Prev in Thread] Current Thread [Next in Thread>