search

panrg / path-properties

A Vocabulary of Path Properties
Other
1 stars 3 forks source link

Change definitions based on feedback #21

Closed renghardt closed 4 years ago

renghardt commented 4 years ago

This PR updates our terminology in line with feedback from IETF 105 and Med's comments. See also issue #1.

A node can now be an entire AS -- can a node also be something like an amorphous collection of path elements in case of partial visibililty? (open question)

A property can now also apply to a set of subpaths -- I wanted this because the sequence of path elements on which a packet is sent can change frequently, even if the endpoints stay the same. Do you agree?

cyrill-k commented 4 years ago

A node can now be an entire AS -- can a node also be something like an amorphous collection of path elements in case of partial visibililty? (open question)

I think defining an abstract node which can be a collection of entities is useful but then the question comes up if we even need links at all or if links are just abstract nodes that only forward packets?

cyrill-k commented 4 years ago

About viewing a link as just another node that forwards packets, this sounds kind of confusing to me and I don't think I agree: I think "forward packets" means different things for nodes and for links - nodes perform routing and switching, links connect one node to another. That's not the same thing, but we don't use "forward" for links in our definition, so we should be fine.

I agree with you. It was just going through my head when I saw the definition of an abstract node. :)

renghardt commented 4 years ago

Thanks for taking another look! So it looks like we can merge this for now?