The MPLS WG Archive

[CATANZARITI Sergio FTR&D/TI <sergio.catanzariti@rd.francetelecom.com>]

Title: RE: MPLS Diffserv Extensions related questions/comments

i certainly see the utility of being able to control
the exp -> phb mapping. i just don't see (yet) that
every phb needs a unique mapping.

Well, let me as an operator to decide if I want to or not that unique mapping. When the ITU-T X/Y will define IP Diff Class Service Classes (ala ITU-T I.362) with a defined PHB <-> service semantics mapping function, then we can reduce our universe of configuration choices. But, at that point I will be either dead or retired in Sicily enjoying beaches and pure Olive Oil :-). The latter choice appeals me better, indeed.

yes. say you could establish several exp -> phb
mappings in your lsrs and each is given an index
unique within the domain. you could signal the
index per lsp. would not that give you the
flexibility you are looking for? how many such
mappings would you need?

I am not questioning the normalization aspect of the mapping, it is that really all depends on what we want to achieve and how to implement an LSP-supported differentiated services offerings environment.

Sergio

--------------------------------------------------------------------
Sergio Catanzariti
Senior Project Manager, Technology Integration
France Telecom R&D
1000 Marina Boulevard Suite 300
Brisbane CA 94005
Tel. 650-875-1526
Fax. 650-875-1505
email:sergio.catanzariti@rd.francetelecom.com
--------------------------------------------------------------------




-----Original Message-----
From:   tom worster [SMTP:tom@ennovatenetworks.com]
Sent:   Tuesday, June 06, 2000 5:52 AM
To:     'CATANZARITI Sergio FTR&D/TI'; curtis@avici.com
Cc:     mpls@UU.NET
Subject:        RE: MPLS Diffserv Extensions related questions/comments

From: owner-mpls@UU.NET [mailto:owner-mpls@UU.NET]On Behalf Of CATANZARITI
Sergio FTR&D/TI
>
> I agree that it would be actually very helpful. I
> would like to build topology-driven LSPs and use
> EXP-PHBs stuff for differentiating CoS-like traffic.

i certainly see the utility of being able to control
the exp -> phb mapping. i just don't see (yet) that
every phb needs a unique mapping.




> I do not understand the "hardware cost friendly"
> problem.

the cost i'm referring to is that a system needs to
provide memory to store an exp -> phb mapping for
every lsp. in a fast implementation that may come
at a noticeable marginal hw memory cost.




> The real issue here is to have a flexible
> LSPs management/control plane (I do not understand
> where the signalling problem is in all this) that
> would allow me to do it in a simple, and semantically
> congruous manner.

yes. say you could establish several exp -> phb
mappings in your lsrs and each is given an index
unique within the domain. you could signal the
index per lsp. would not that give you the
flexibility you are looking for? how many such
mappings would you need?




> It should not be that hard, at least at the
> provisioning level, and I guess I saw it already but
> I do not recall now which vendor was it :-).

it's certainly doable. but it comes at a cost in
some designs. in designs in which the memory for
each lsp is reallocated, providing support for
random signalled mappings comes at a cost.

c u
tom