Hi Linda,

> 
> Your draft is written very well and clear.

Thank you :-)

>  
> Here are some questions to the draft:
>  
> -         Figure 3 (Page 8): Is the  "nexthop-list"  counted as the "Action"
> when the "match" condition is met?


YesŠ.

>  
> -         What is the relationship between "nexthop-list" and
> "route-attribute"?


Route-attributes are properties of the route. Nexthop-list is your
action-list. So they are kind of orthogonal. For ex. Route-preference allows
route selection across multiple similar routes. Only the route that gets
selected will have it¹s nexthop-list applied when a packet hits that route.

> 
> 
>  
> 
> -         Why the Route-attribute ranges 0-N, whereas the ³nexthop-list²
> starts from 1? 
> 
>  

Route-attributes are optional entities. Whereas there MUST be at least one
next-hop for the given match condition. What¹s the point of a route without
a nexthop. If you want to ³discard² a packet by the given route, ³discard²
is a next-hop.


> -         What kind of scenario will cause ³Lists of lists², i.e. the ³Chain²
> in the ³nexthop-list-member²?
> 
>  
> 
> You have a statement in the draft:
>  
> Isn¹t one node supposed to deal with only the outer header?
>  

In a simple case, one node will deal with only the outer header. However if
a node is originating a tunneled packet, it might need to send 2 headers
over. Other examples include a PW - MPLS over some transport (MPLS or GRE
for instance) and VxLAN over IP.

draft-bitar-i2rs-service-chaining and draft-hu-i2rs-overlay-use-case could
likely use this paradigm as well.

Thanks
Nitin


_______________________________________________
i2rs mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/i2rs

Reply via email to