[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: FW: Questions about tunneltrace
The following is a reponse to both of your postings. Comments inline.....
> -----Original Message-----
> From: Eric Rosen [mailto:firstname.lastname@example.org]
> Sent: Friday, March 02, 2001 11:24 AM
> To: Ron Bonica
> Cc: email@example.com
> Subject: Re: FW: Questions about tunneltrace
> Ron> Strictly speaking, I am not tracing through the
> user-plane, but
> Ron> bouncing back and forth between the user-plane and the
> control-plane. I
> Ron> have no choice but to do so.
> Perhaps. But then your requirement of not relying on the
> control plane
> needs to be removed.
Point taken. I will remove it in the next version of the requirements draft.
> Once that requirement is removed, though, various other
> alternatives suggest
> themselves, such as using SNMP to get the information from
> the MIBs, or
> having a special "trace packet" that accumulates the path
> information inside
> itself, and then returns the entire trace in one packet back
> to the user.
> It would be useful to see some discussion of why the proposal
> in the draft
> is better than these alternatives. (Note that I'm not saying
> that it isn't
> better than these alternatives.)
Again, point taken. I will add a section concerning alternative approaches
in the next version of the protocol specification.
> Ron> Isn't that the kind of layer violation that we are trying to avoid?
> Can you articulate this layer violation?
I will try. Somebody please jump in if I misspeak myself.
We are trying to avoid the trap into which draft-ietf-mpls-icmp fell. The
basic problem with that draft was that it looked inside an MPLS payload,
found and IP header, and assumed that the source address specified by that
IP header was meaningful within the LSRs layer three routing context.
For example, lets say that the source address is 10.1.1.1 and the LSR has a
route to 10.1.1.1. Is the host to which the LSR has a route the same one the
originated the datagram? If the LSR does not have a route to 10.1.1.1, it
sends the ICMP message to the egress LSR. Is the host to which the egress
LSR has a route the same one the originated the datagram?
GTTP dances around this issue by relaying the response through the head end
of each tunnel in the stack. But the price that it pays is to bounce back
and forth between the data and control planes.
(I hope that I haven't mangled the argument too badly.)