[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: About LSP Stitching
Thanks for replying.
Below are the two main assumptions on which my query is based.
Plz correct me if I am wrong.
1. E2e LSP & LSP segment are two independant LSP's.
2. TE-Link is a kind of abstraction.
User of TE-Link should not know about the implementation details of the TE-Link.
Whether it is a FA or a LSP segment as a TE-Link, an LSP setting up over these TE-Link should not be aware of the TE-Link setup details(whether its a FA or a LSP segment or a Physical Interface or others).
So if we go by this, the below example which i gave holds true.
Plz let me know your view on this.
Thanks and regards,
----- Original Message -----
From: Arthi Ayyangar <email@example.com>
Date: Saturday, April 9, 2005 4:22 am
Subject: Re: About LSP Stiching
> > There is no way for LSP segment head node, to identify that a
> > Outif selected is a LSP Segment for setting up E2e LSP.
> ---> What do you mean ? Why would the head-end of an LSP segment
> not know
> that selected nexthop is the LSP segment ? It is the head-end of
> the LSP
> segment selecting what you call Outif.
> > That is Some extension will be needed to RSVP that will allow RSVP
> > message(for E2e LSP) processing at LSP segment head node to be
> processed> according to the Sticthing needs.
> ------> I don't think there is any need for any more
> extensions..see my
> explanation above and below.
> > For Eg: LSP1-2 is an E2e LSP setup over an LSP segment LSPA-B.
> > 1-----A--D---E--B-----2
> > Now if A gets the Path message for LSP1-2, and A selects LSP
> segment LSPA-B as outif.
> > Then how does RSVP(at Node A) identify that the outif selected is
> a LSP segment?
> -------> A is doing nexthop (what you call outif) selection, A is the
> head-end of LSP segment. A knows that nexthop is an LSP segment.
> > Also when A receives Resv for LSP1-2(from B), how can it ignore
> > Label in it.
> > There should be a way to determine at node A, that a LSP segment
> > is in use.
> -----> A chose to stitch the e2e LSP to the LSP segment in the
> first place.
> > One possible way could be that Resv message for LSP1-2 should be
> > sent(by B) with some special Label value.
> > This Label value should indicate the node A that the Resv is
> > received for a TE-Link(of type LSP Segment).
> > This will help node A to ignore Label value in Resv msg for E2e
> > Lsp LSP1-2.