The MPLS WG Archive

[Curtis Villamizar <curtis@avici.com>]

In message <4.2.0.58.J.20000329230505.00a41be0@momotaro.labs.nec.co.jp>, Atsush
i Iwata writes:
> Hi, all
> 
> The draft that peter mentioned is our following draft.
> 
> draft-fujita-ospf-te-summary-00.txt
> 
> This approach assumes a hierarchical QoS path computation that PNNI routing 
> protocol does in the ATM world.
> 
> I would like to get any opinion or any suggestions (Pros and Cons) on this 
> approach.
> 
> Regards,


Someone needed to do this.  You are missing some key parameters like
reservable bandwidth at each of the 8 priority levels.  There is no
good way to specify colr since we're not talking about a link.

This is not interprovider, but rather inter-area in OSPF.

Besides advertising this information a LSR also needs to do something
with it when forming an ERO.  You should have some sort of usage
section.  For example, if a BGP route has a next hop covered by the
area, then the ERO would contain a router to an ABR ending with a
loose hop to the BGP next hop address.  The ABR may either replace the
loose hop with a set of strict hops, or put the LSP into an existing
aggregate tunnel.

In this case the aggregate tunnel may go from the ABR all the way to
the network edge.  The PSC in draft-kompella-mpls-optical-00.txt
serves a similar purpose when the endpoints of the aggregate tunnel
are not the ABR and the area edge.  In the former case there is no
signaling of a PSC required within the area.

Curtis