Spring WG:

On July 12, I asked as a WG chair for an in-depth review of
draft-ietf-idr-sr-policy-nrp-03 from Spring in the following email:

https://mailarchive.ietf.org/arch/msg/spring/Hohar_8E8wEpTgWJQguyXTRllf4/

The information on the mail list had a discussion between Jie Dong and 
Zhengqiang Li.
They discussed whether NRP should be decoupled from SRP Policy.

Jie Dong stated:
"SR Policy can be considered as the SR instantiation of TE tunnels ...
an SR Policy can be enhanced as SR-TE tunnels with resource guarantee.
>From this perspective, it would be better to associate an SR Policy with one 
>NRP.
That said, as SR segments can be shared by multiple SR Policies, it is possible
to share a common path among multiple resource guaranteed SR Policies by using 
the same SID list."

Zhengqiang Li replied:
"If one SR Policy is only for one NRP, I don't think we need any slice id for 
the SR policy,
just steer the specific traffic to the policy, the isolation QoS
will be satisfied by the SR policy since it is built only for the specific NRP."

Jie to Zhengqiang:
"As I mentioned earlier, the SR SIDs can be shared among multiple SID lists,
and the SID lists can be shared among multiple SR Policies,
while each SR Policy is associated with a specific intent.
For service traffic which require(s) both explicit path and guaranteed
bandwidth (as the service intent), the SR Policy used for traffic steering
needs to provide both characteristics, hence it needs to be a resource 
guaranteed
SR Policy which can meet the intent. This way, the SR Policy
service steering mechanism can be kept unchanged."

Zhengqiang replied to Jie:
"In the approach you proposed, the number of SR Policies in the network
will increase dramatically, imposing an unaffordable maintenance burden.
This issue should be noted and considered by both IDR and SPRING.
In my opinion, SR Policy is the road between two nodes and the NRPs or
network slices are the lanes, such as a carpool lane, in the road."

The discussion did not reach a conclusion.  As the IDR Shepherd,
I need to know:
1. Is there any problem with service traffic requiring both explicit
Path and guaranteed bandwidth (for service intent)?

2. Does NRP help or hinder maintenance burden? Why and Why not?

3. Is there a scaling issue that Spring is concerned about with the
NRP approach?
4. Does the Spring WG have any other concerns about publishing 
draft-ietf-idr-sr-policy-nrp?

Cheerily, Sue Hares



This discussion did not come to a common conclusion.
_______________________________________________
spring mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to