The MPLS WG Archive
[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index]
Decreasing TE-LSP bandwidth, "interoperability issue" (fwd)
-
From: "David Allan" <dallan@nortelnetworks.com>
-
Date: Wed, 6 Nov 2002 08:50:03 -0500
-
Cc: Der-Hwa Gan <dhg@juniper.net>, "Gray, Eric" <egray@celoxnetworks.com>, mpls@UU.NET, swallow@cisco.com
Title: RE: Decreasing TE-LSP bandwidth, "interoperability issue" (fwd)
Hi George:
I was envisioning a case where a bandwidth change was not a separate tunnel setup, the LSP ID changed without any interruption or change in the conntectivity of the path . If there is build the new then tear down the old, then there is no gap in any monitoring case.
As for TTSI, it's simply a node specific LSP Identifier that allows for an IPv4/v6 LSR ID and 32 bits for an LSP identifier. The assumption is that the originating LSR most usefully administers LSP-IDs independently of the individual control protocol used. The initial draft of Y.1711 was focused on P2P ER-LSPs which meant RSVP-TE or CR-LDP, hence the stipulation that the first two bytes of the LSP-ID were padded with zeros to align with LSP-ID as defined in the signalling protocols. If there is a problem with this, it is not leaping out at me....
cheers
Dave
> -----Original Message-----
> From: George Swallow [mailto:swallow@cisco.com]
> Sent: Wednesday, November 06, 2002 8:37 AM
> To: Allan, David [CAR:NS00:EXCH]
> Cc: Der-Hwa Gan; Gray, Eric; mpls@UU.NET; swallow@cisco.com
> Subject: Re: Decreasing TE-LSP bandwidth, "interoperability
> issue" (fwd)
>
>
>
> > Only scenario that comes to my mind is that some form of
> data plane probing
> > of the LSP is going on (be it LSP-PING, Y.1711 or
> whatever). Changing the ID
> > means there is a small window in which it may appear as a
> transient fault to
> > the probing system.
>
> With LSP ping, you would ping the first LSP until the second is set
> up. Once that is complete you can start pinging it as well. You can
> even wait for a successful reply before ceasing pinging the first and
> tearing it down. (Though I think that's probably overkill.) So you
> don't have a gap, you actually have an overlap.
>
> With Y.1711 I don't understand how you would map an RSVP tunnel/lsp-id
> to a TTSI. There are just enough bits to fit both RSVP fields into
> the TTSI LSP-ID, but that doesn't leave anying for LDP, BGP, etc.
> Maybe you can enlighten us.
>
> ...George
>
>
>
> ==================================================================
> George Swallow Cisco Systems (978) 497-8143
> 250 Apollo Drive
> Chelmsford, Ma 01824
>
| |
|