[c-nsp] qos plan - advice please

[Robert Blayzor]

On Aug 30, 2013, at 12:20 PM, Mark Tinka <mark.tinka at seacom.mu> wrote:
> I've always supported DiffServ. I've found using RSVP to 
> signal admission control to be otherwise heavy (there was a 
> reason it never took off in the first place, despite how 
> noble the idea was).


I'd have to agree.  Especially if you're building a greenfield network here.  Not like QoS policies can't be done later on; but at least you can test them without adversely effecting traffic.  

IntServ really isn't practical unless you you're going to do strict admission control and make sure bandwidth is completely reserved on every link, even in failure scenarios.  It's not really practical, scalable or easy to manage in modern networks.

It doesn't take that much time to make up some service levels and get your transit link policies in place.  When  you start getting into lots of TE tunnels and fast-reroute, etc.. that's when they get a bit more complex and time consuming.

-- 
Robert Blayzor
INOC, LLC
rblayzor at inoc.net
http://www.inoc.net/~rblayzor/