Hi Pushpasis,

On Jul 30, 2015, at 11:11 AM, Pushpasis Sarkar 
<[email protected]<mailto:[email protected]>> wrote:

Hi Acee,

From: "Acee Lindem (acee)" <[email protected]<mailto:[email protected]>>
Date: Thursday, July 30, 2015 at 8:37 PM
To: Pushpasis Sarkar <[email protected]<mailto:[email protected]>>
Cc: Shraddha Hegde <[email protected]<mailto:[email protected]>>, "Les 
Ginsberg (ginsberg)" <[email protected]<mailto:[email protected]>>, Robert 
Raszuk <[email protected]<mailto:[email protected]>>, "Aissaoui, Mustapha 
(Mustapha)" 
<[email protected]<mailto:[email protected]>>,
 Stephane Litkowski 
<[email protected]<mailto:[email protected]>>, 
"[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>, Uma Chunduri 
<[email protected]<mailto:[email protected]>>, Hannes Gredler 
<[email protected]<mailto:[email protected]>>
Subject: Re: [spring] Modeling SRGB configuration for draft-ietf-spring-sr-yang

Hi Pushpasis,

On Jul 30, 2015, at 10:54 AM, Pushpasis Sarkar 
<[email protected]<mailto:[email protected]>> wrote:

Hi Acee,

From: "Acee Lindem (acee)" <[email protected]<mailto:[email protected]>>
Date: Thursday, July 30, 2015 at 7:18 PM
To: Pushpasis Sarkar <[email protected]<mailto:[email protected]>>
Cc: Shraddha Hegde <[email protected]<mailto:[email protected]>>, "Les 
Ginsberg (ginsberg)" <[email protected]<mailto:[email protected]>>, Robert 
Raszuk <[email protected]<mailto:[email protected]>>, "Aissaoui, Mustapha 
(Mustapha)" 
<[email protected]<mailto:[email protected]>>,
 Stephane Litkowski 
<[email protected]<mailto:[email protected]>>, 
"[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>, Uma Chunduri 
<[email protected]<mailto:[email protected]>>, Hannes Gredler 
<[email protected]<mailto:[email protected]>>
Subject: Re: [spring] Modeling SRGB configuration for draft-ietf-spring-sr-yang

Robert, Pushpasis, Shraddha,

I guess I don’t understand why this gives you any operational advantage when 
you are migrating between protocol (e.g., from IS-IS to OSPF ;^). In fact, if 
you are using separate label spaces it will make migration more difficult and 
you will have to migrate your routing domain in shot rather than incrementally.
[Pushpasis] All routers will program both transit mpls labels and create two 
parallel MPLS data planes well before the last step when all the ingress router 
will be instructed to switch traffic onto the new MPLS data plane (programmed 
by the new protocol) by changing the admin distance. That way the migration 
shall become seamless. And then once all ingreess routers has been swicthed to 
the new protocol and new MPLS dataplane, the old protocol and old MPLS data 
plane can be just removed from all the routers. Hope it is clear now…

Normally only the protocol with the lower preference or admin distance would 
install a route to a particular prefix (including the label) into the RIB/LIB 
for the base IP unicast topology. I guess you are envisioning separate RIBs per 
protocol as well?
[Pushpasis] What you are saying is true for the IP/V6—>MPLS route in RIB/FIB on 
ingress routers. It is NOT TRUE for MPLS transit routes in LFIB. In MPLS 
architecture no two protocols installs the same label in LFIB. Each protocol 
(LDP/RSVP) uses different transit labels for the same destination IP/V6 FEC.

That is one way to implement it but it makes more sense for the IGP label 
imposition and label stitching to both be dependent on the route being 
preferred.
However, in another conversation I have been convinced that the network 
consolidation use case could benefit from configuration of per-protocol SRGB so 
I’m no longer opposed to this being a feature of the YANG model.

Thanks,
Acee





Thanks,
Acee





Thanks,
Acee

On Jul 30, 2015, at 3:11 AM, Pushpasis Sarkar 
<[email protected]<mailto:[email protected]>> wrote:

+1

From: Shraddha Hegde <[email protected]<mailto:[email protected]>>
Date: Thursday, July 30, 2015 at 8:58 AM
To: "Les Ginsberg (ginsberg)" <[email protected]<mailto:[email protected]>>, 
Robert Raszuk <[email protected]<mailto:[email protected]>>
Cc: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>, Uma 
Chunduri <[email protected]<mailto:[email protected]>>, 
"Aissaoui, Mustapha (Mustapha)" 
<[email protected]<mailto:[email protected]>>,
 "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>, Pushpasis Sarkar 
<[email protected]<mailto:[email protected]>>, Hannes Gredler 
<[email protected]<mailto:[email protected]>>
Subject: RE: [spring] Modeling SRGB configuration for draft-ietf-spring-sr-yang


[Shraddha] route preference is local to router and the neighbor never knows 
what the other is going to choose.  If there are inconsistencies in two 
protocols, there are going to be loops even when they use same SRGB. Keeping 
different labels makes it easier to troubleshoot.

_______________________________________________
spring mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/spring



_______________________________________________
spring mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/spring

Reply via email to