The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2000-Jul> msg00420



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Two last calls (draft-ietf-mpls-diff-ext-06.txt)

  • From: Shahram Davari <Shahram_Davari@pmc-sierra.com>
  • Date: Thu, 27 Jul 2000 08:09:46 -0700
  • Cc: Francois Le Faucheur <flefauch@cisco.com>, Shahram Davari <Shahram_Davari@pmc-sierra.com>, mpls@UU.NET

Hi Lou,

I like Francois' proposal. Your proposal forces all the new Diffserv-LSR
boxes to implement pre-Diff-ext (non-DS) mechanisms (which is obsolete).
This will cause the pre-Diff-ext mechanisms which are non standard, to drag
for ever, even after all the LSRs in the world are using Diffserv.

I think Francois' proposal fulfills your concern. Because some vendors may
opt to implement both DS and non-DS in their boxes (LSRs), and those
operators that are concerned with upgrading their existing non-DS equipment
could buy them, while some other vendors may just implement DS LSRs, and the
operators that didn't have non-DS LSRs will buy them and won't pay extra for
something that they don't need.

Francois' proposal also gives the non-DS operators an option of upgrading
all their LERs to the (DS + non-DS) LERs, and after they have done so they
don't need to upgrade their core LSRs to (DS+ non-DS), rather they could
upgrade their core LSRs directly to DS LSRs and as I mentioned in my last
email this would work perfectly because the LERs could set both EXP and COS
field. The EXP could be used by DS-LSRs and the COS could be used by non-DS
LSRs.

Regards,
-Shahram 

>-----Original Message-----
>From: Lou Berger [mailto:lberger@labn.net]
>Sent: Thursday, July 27, 2000 9:49 AM
>To: Francois Le Faucheur
>Cc: Lou Berger; Francois Le Faucheur; Shahram Davari; mpls@UU.NET
>Subject: RE: Two last calls (draft-ietf-mpls-diff-ext-06.txt)
>
>
>Francois,
>
>At 05:45 PM 7/26/00 +0200, Francois Le Faucheur wrote:
>>Lou,
>>
>>At 12:04 24/07/2000 -0400, Lou Berger wrote:
>>
>> >
>> >How about including an "SHOULD" level option that allows a 
>MPLS-DS LSR to
>> >treat LSPs signaled without the DIFFSERV object as non-DS 
>LSPs?  I think
>> >this would address this issue.
>> >
>>
>>Just to be sure I understand the proposal:
>>The text would say that when there is RSVP signaling without 
>the Diff-Serv 
>>object, the LSR SHOULD (instead of MUST) interpret this as a 
>request for 
>>an "E-LSP using Preconfigured EXP<-->PHB Mapping".
>>(which would imply that if there's a really good reason to do 
>otherwise - 
>>eg. to maintain some pre-Diff-Ext QoS support - then you could do so).
>>
>>If that's the proposal, that works for me.
>
>On additional recommendation and we're there. Something like:
>An implementation SHOULD include an option that overrides 
>this.  When this 
>"override option" is set, the LSR then treats LSPs signaled 
>without the 
>DiffServ object as standard, non-DS, LSPs.
>
>How does something like this sound?
>
>> >(BTW which heterogeneity and transition issues are you 
>addressing with the
>> >objects proposed in your new draft?)
>> >
>>
>>The new draft focuses on Diff-Serv-aware Traffic Engineering whereby 
>>different bandwidth constraints can be applied to different Diff-Serv 
>>Classes. This involves advertising more than one "unreserved 
>bandwidth" 
>>information in IGPs and using the appropriate one for path 
>computation for 
>>each Diff-Serv Class. Also, it involves performing admission 
>control over 
>>different bandwidth pools at LSP set-up.
>>The transition situation we focused on is when migrating from 
>existing TE 
>>(which we refer to as "Aggregate TE") to "Diff-Serv-awrae-TE". More 
>>precisely , the migration situation involves a mix of:
>>         - "old" LSRs which are Diff-Serv capable and (Aggregate) TE 
>> capable (ie support a single bandwidth constraint, aka 
>single bandwidth 
>> pool, aka compute the same route regardless of the Diff-Serv 
>Class transported)
>>         - "new" LSRs which are Diff-Serv capable and 
>Diff-Serv-aware-TE 
>> capable (ie support a multiple bandwidth constraints, aka multiple 
>> bandwidth pools, aka may compute different routes depending on the 
>> Diff-Serv Class transported).
>>
>>Note that we are not proposing new objects just to address 
>this transition 
>>issue. Rather we are saying that the new objects/procedures that get 
>>defined for Diff-Serv-aware-TE MUST satisfy this migration goal.
>>
>>Hope that clarifies
>
>Sounds like parallel issues to me.  But it doesn't matter as 
>long as we 
>agree on the above recommendations, i.e., I think we have a 
>compromise that 
>covers
>the open issue.
>
>Thanks,
>Lou
>
>>thanks
>>
>>Francois
>>
>> >Thanks,
>> >Lou
>>_________________________________________________________________
>>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
>>_________________________________________________________________
>