The MPLS WG Archive

[Francois Le Faucheur <flefauch@cisco.com>]

Hi,

We've just posted <draft-ietf-mpls-diff-ext-05.txt>. It should be available on the IETF server shortly.

This is an update on 04.txt. The main point is the resolution of the single issue left open by 04.txt (after the "informal last call) which is details ingress/egress Diff-Serv behavior when pushing and popping labels (including hierarchical operations). The principles of the resolution for this have already been described on the alias in March (see message below).

Other changes are relatively minor:
	- clean up of mappings in differnet cases involving ATM/FR (eg LC or non LC interfaces)
	- typos/clarifications

As indicated earlier, we will ask the WG chairs to issue an official WG Last Call.

Francois


>X-Sender: flefauch@europe.cisco.com
>X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0.2 
>Date: Thu, 23 Mar 2000 19:10:14 +0100
>To: mpls@UU.NET
>From: Francois Le Faucheur <flefauch@cisco.com>
>Subject: Proposal for last open issue re draft-ietf-mpls-diff-ext-04.txt
>Cc: flefauch@europe.cisco.com, bsd@cisco.com, flefauch@cisco.com,
>        liwwu@europe.cisco.com, pasi.vaananen@ntc.nokia.com,
>        ram@nexabit.com ('Ram Krishnan'),
>        Shahram_Davari@pmc-sierra.com (Shahram Davari),
>        Pierrick.Cheval@alcatel.fr, jh@lohi.eng.telia.fi
>Sender: owner-mpls@UU.NET
>X-SMTP-HELO: wodc7-1.corprelay.mail.uu.net
>X-SMTP-MAIL-FROM: owner-mpls@UU.NET
>X-SMAP-Received-From: outside
>X-SMTP-PEER-INFO: wodc7-1.corprelay.mail.uu.net [192.48.96.68]
>
>Hello,
>
>A single issue from the Informal Last Call on
>draft-ietf-mpls-diff-ext-03.txt had not been closed as of
>draft-ietf-mpls-diff-ext-04.txt. This is the detailed ingress and egress
>Diff-Serv behavior when pushing and popping labels.
>
>Here is a high level proposal for resolution of this issue.
>
>The proposal borrows heavily from draft-ietf-diffserv-tunnels-00.txt,
>'Differentiated Services and Tunnels' from D Black.
>
>Most of the characteristics of the problem statement addressed by this
>draft are common with the issue we need to address in
>draft-ietf-mpls-diff-ext-04.txt:
>	- intermediate nodes only see and operate on the "outer" Diff-Serv
>information
>	- IP tunnels/LSPs Tunnels are unidirectionnal
>	- the outer Diff-SErv information may get modified somewhere on the span
>of the IP Tunnel/LSP Tunnel
>
>There is a few notable differences between an LSP Tunnel and an IP Tunnel
>though:
>	- Penultimate Hop Popping (PHP) results in the LSP Tunnel Diff-SErv
>information not being transmitted to the LSP Tunnel Egress.
>	- two level LSP Tunnels from same ingress to same egress are expected to
>be common occurence in MPLS-land for MPLS VPN so we need to address this
>case. I have only covered below the case of unlabelled packets going into a
>two level LSP tunnel because MPLS VPN is the only case I can think of for
>this two level LSP tunnels.
>
>Based on this, I would propose that:
>
>	- like with draft-ietf-mpls-diff-ext-04.txt, we recognise that there are
>two meaningful conceptual models which are "Uniform" model and "Tunnel" model.
>
>	- like with draft-ietf-mpls-diff-ext-04.txt, we recognise that they are
>both useful.
>
>	- unlike with draft-ietf-mpls-diff-ext-04.txt, we can play with PHP to
>simplify operations of the egress in case of single level LSP Tunnel
>
>	- for single level LSP Tunnel, to achieve "tunnel" mode:
>		* Ingress: Egress PHB is reflected in pushed label. Egress PHB is NOT
>reflected in swapped label or encapsulated IP DSCP.
>		* Egress: requests PHP and therefore use "outer" Diff-Serv info for
>incoming PHB determination (ie outer label for received labelled packet or
>DSCP for received unlabelled packet).
>
>	- for single level LSP Tunnel, to achieve "uniform" mode:
>		* Ingress: Egress PHB is reflected in pushed label. It does not matter if
>Egress PHB is also reflected in swapped label or encapsulated IP DSCP since
>this will be rewritten at egress ayway.
>		* Egress: requests No-PHP and use "outer" Diff-Serv info for incoming PHB
>determination (ie outer label before POP for received labelled packet, or
>DSCP for received unlabelled packet). 
>
>	- for two level LSP Tunnel (MPLS VPN), to achieve "tunnel" mode:
>		* Ingress: Egress PHB is reflected in both pushed labels. Egress PHB is
>NOT reflected in encapsulated IP DSCP.
>		* Egress: requests PHP but still can NOT use "outer" label for incoming
>PHB determination. Egress must be able to use the DSCP of the packet (as
>exposed after the POP) to do PHB determination. In other words, Egress must
>be able to (logically) do the PHB determination after the POP.
>
>	- for two level LSP Tunnel (MPLS VPN), to achieve "uniform" mode:
>		* Ingress: Egress PHB is reflected in both pushed labels. It does not
>matter if it is also reflected in encapsulated IP DSCP since this will be
>rewritten at egress ayway.
>		* Egress: requests PHP anyway (because with MPLS VPNs both labels are
>expected to reflect the same PHB) and use "outer" label/or DSCP for
>incoming PHB determination (before doing POP).In other words, Egress must
>be able to (logically) do the PHB determination before the POP.
>
>
>Thanks for comments. 
>As soon as we've reached agreement I will proposed detailed text for the
>next rev of the draft.
>
>Francois
>
>_________________________________________________________________
>Francois Le Faucheur   
>Development Engineer, IOS Layer 3 Services 
>Cisco Systems 
>Office Phone:   	+33 4 92 96 75 64
>Home Office Phone:     +33 4 92 94 00 78
>Mobile :               +33 6 89 108 159
>Vmail:                 +33 1 58 04 62 66
>Fax:                   +33 4 92 96 79 08
>Email:          	flefauch@cisco.com
>_________________________________________________________________
>Petra B - Les Lucioles - 291, rue Albert Caquot - 06560 Valbonne - France
>_________________________________________________________________ 
>