The MPLS WG Archive

[neil.2.harrison@bt.com]

Sounds fine to me Zhi.  I keep trying to make the point that availability
SLAs and QoS SLAs are quite different animals....the latter only has
relevance when a connection (eg an LSP) is in the up-state.  So I need to be
able to know the former before I can do the latter.  This is indeed why I
also have a problem with the 'DiffServ' only approach which mutates failures
into QoS hits, ie the forwarding behaviour of a traffic type (eg voice that
needs EF) bears no relationship to its survivability requirements.  I want
to be able to offer my VPN customers different avail SLAs and this needs
some notion of hard BW partitioning and overall VPN topology survival (vs
other peer, different customer, VPNs)....if I merge all their traffic, then
how can I do this?

Of course this (ie hard BW partitioning) is forced once we start using
CO/cct-sw fabrics like SDH and the OTN as 'LSPs' (I am not getting into the
common control-plane issue here!....but you know my views on this by now).
But the *principle* still applies everywhere else.

Bottom-line.....I am only interested in QoS once I know the availability
issue is cracked.  To crack availability all defects, and there correct
(user-plane) handling, must be defined.  We are not there yet.

Neil

> -----Original Message-----
> From:	Zhi-Wei Lin [SMTP:zwlin@lucent.com]
> Sent:	Monday, October 23, 2000 5:17 PM
> To:	Sid Chaudhuri
> Cc:	'Kireeti Kompella'; xuyg@lucent.com; yxue@UU.NET;
> ip-optical@lists.bell-labs.com; mpls@UU.NET
> Subject:	Re: [IP-Optical] RE: Optical link bundling. Was Re:
> DraftMinutes From Pittsburgh
> 
> Hi Sid,
> 
> And to generalize even further, if a service provider sets up three
> grades of service (e.g., bronze, gold, platinum), the protection
> information should already be embedded within those service grades. For
> example, bronze is no protection at all, gold is dynamic mesh
> restoration, and platinum is 1+1 protection. 
> 
> I think what is most important to a service provider and their customers
> are the availability of the connection, i.e., is the connection up
> 99.99% or 99.999% or some other number. How the service provider chooses
> to handle how to meet that availability number is up to the service
> provider and their TE.
> 
> Am I mis-representing the service provider here? Maybe some service
> providers can comment on whether the above description sounds right???
> 
> Thanks
> Zhi
> 
>