Re: [CCAMP] [Technical Errata Reported] RFC4872 (3304)

["Zafar Ali (zali)" <[e-mail]>]

Hi Adrian- 

Please see in-line. 

Thanks

Regards ... Zafar 


> -----Original Message-----
> From: Adrian Farrel [mailto:[e-mail]]
> Sent: Thursday, August 02, 2012 4:56 PM
> To: Zafar Ali (zali); [e-mail]
> Cc: [e-mail]; [e-mail];
> [e-mail]
> Subject: RE: [CCAMP] [Technical Errata Reported] RFC4872 (3304)
> 
> Zafar,
> 
> [snip]
> 
> > > But the general paradigm here is that you are willing to use the
> best
> > > available LSP when it is set up in the first place, the best
> available LSP
> > > when you re-route after failure, and the best available LSP when you
> > > "revert".
> >
> > The original paths in the network may be explicit (with mix of dynamic
> for
> other
> > services). In this case, service of a LSP may not come back to the
> original
> explicit
> > paths if some other (dynamic) LSP "stole" the resources for the failed
> working
> > LSP (if resources are released during a failure) - as you noted above.
> 
> Hmmm.
> 
> You are saying that when the initial LSP is established it is OK to use
> and ERO
> to "force" it down the path you want. You are not saying that the
> network
> resources must be pre-reserved for the initial LSP.
> 
> Why is it not good enough to use an ERO to "force" reversion?
> 

I did not understand what you meant by "use an ERO to "force" reversion". The issue in my mind is as follows: 

1. We are at time t0. 
2. Working LSP is setup on THE nominal path (P) with setup/ holder priority (s,h). For simplicity assume P is explicitly configured by SP. 
3. At time t1 > t0, a resource, say link L fails, along the nominal Path (P). 
4. Ingress tears down working LSP. 
5. A restore LSP or protecting LSP starts to carry traffic. 
6. Link L comes back up at t2 > t1. 
7. Some other LSP(s) with setup/ hold priority better than working LSP setup/ holder priority (s,h) claims resources from L. 
8. Nominal Path P is no longer available for working LSP and this is the only Path SP wants the working LSP to take. 
9. Reversion cannot be applied as working LSP stays down. 

Thanks

Regards... Zafar 

> [BTW: the discussion of whether you might want to do it (that we are
> having now)
> is orthogonal to whether it is possible. I have no issue that it is
> possible,
> and I don't think any changes are needed to any documents to enable it
> or to say
> that it is possible.]
> 
> Adrian