RE: I-D ACTION:draft-andersson-mpls-g-chng-proc-00.txt

[neil.2.harrison@bt.com]

Thanks Gert....I have tried to answer you below.  regards, Neil

Gert Grammel wrote 06 March 2003 19:57
>   1. I never understood why you were so emotional on overlay 
> models as compared
>      to peer but at the same time advocate to use PNNI (Peer 
> Network to Network
>      Interface) as an alternative.
NH=> The L1/0 networks have multiple clients....so there is no peering with
any of them in the control-plane sense.  For external NNIs to a different
operator within the *same* L1 network one would invariably use static
routings.  This is irrespective of the actual choice of signalling protocol.

>   2. The addressing issue you've raised is related to UNI 
> which was so far
>      (until recently) not part of the CCAMP work. So I don't 
> see here that you
>      were forced to use IPv4 Addresses.
NH=> It would make more sense to introduce V6 here IMO....the point on V4 is
not to start with a limited address space if there is no reason to.

>   3. About RSVP vs. PNNI I am not religious, why not using 
> SS7? In any case I
>      don't think that the IETF is the right place to discuss 
> on PNNI. Honestly I
>      don't know which one is 'better' but I don't believe 
> that it is always the
>      best protocol that will win at the end. If you have one 
> why do you need yet
>      another one (and in IETF we had already two ;-)?
NH=> There is more operational experience of pnni and in the opinion of our
signalling experts its a better choice than RSVP.

>   4. I don't see your point on running a routing protocol on 
> L1/0 why no just
>      using RSVP-TE signaling and omit OSPF?
NH=> You could do this I guess. 

>   5. About the access to centralized databases and management 
> system I believe a
>      solution could be found once the specific aplication and 
> communications
>      requrements are clear. I don't see GMPLS excluding this at all.
NH=> True.  However, hope you understand why we see the NMS as being *the*
overriding reference.
> 
> As an observation I'd like to say that all your arguments are 
> equally valid
> applicable to any Layer technology and not specific to L1/0. 
NH=> They don't apply to a cnls mode.  But I agree many of the aspects could
apply to a co pkt-sw mode.
<snipped to end>