Re: [mpls] Some clarifications on draft-dasmith-mpls-ip-options-01.txt

["David Smith (djsmith)" <djsmith at cisco.com>]

Hi Rajeev,

Feedback as follows:

(1) Where exactly label imposition occurs on the ingress LER (fastpath
vs. slowpath) is really out of scope. Ingress LERs need only to support
MPLS encapsulation of packets with IP option fields, and not necessarily
MPLS encapsulation within a specific LER forwarding path (fastpath vs.
slowpath). While the various security risks of IP option pkts against
LSRs can be mitigated using MPLS encapsulation at the LER, MPLS
encapsulation does not protect an LER including its slowpath since it
receives IP option pkts as they enter the MPLS domain. How an ingress
LER processes IP header option fields "before" MPLS encapsulation is out
of scope as it is not relevant to MPLS. The draft only provides LER
rules for supporting MPLS "encapsulation" of IP option packets. 

(2) Same applies here as stated above. How an LER processes the IP
options fields after MPLS decapsulation is out of scope, and has nothing
to do with MPLS. 

Regarding the default behavior, we'll change the text to "....and the
policy SHOULD be enabled by default". This should provide enough
flexibility. We'll post the WG version within a week.

Thanks for the feedback.

Regards,

/dave


-----Original Message-----
From: mpls-bounces at ietf.org [mailto:mpls-bounces at ietf.org] On Behalf Of
Rajeev Manur
Sent: Wednesday, November 26, 2008 7:12 PM
To: mpls at ietf.org
Subject: [mpls] Some clarifications on
draft-dasmith-mpls-ip-options-01.txt


Hi Authors,

I understand the overall intent of this draft and I support it. I just
wanted to get some clarifications on the following cases,

(1) I understand that you propose the ingress LERs to impose a label
stack on the received IP_OPTIONs pkts to prevent these pkts from hitting
the slow path on the downstream LSRs. It is not clear if you expect this
label imposition on ingress LER to happen in the fastpath or slow path?

(2) Similarly it would be good to clarify if such packets need to be
handled in slow-path / fast-path once they reach the egress of the LSP
tunnel.

In my opinion this draft does not have to mandate the default behavior.
We could just specify label imposition of IP_OPTIONs pkts as one of the
mandatory modes of operation for compliance, letting the
operators/vendors chose the default that suits them best.

Thanks !

With regards,
Rajeev
_______________________________________________
mpls mailing list
mpls at ietf.org
https://www.ietf.org/mailman/listinfo/mpls
_______________________________________________
mpls mailing list
mpls at ietf.org
https://www.ietf.org/mailman/listinfo/mpls