Re: [lisp] draft-ietf-lisp-introduction - Design Principles and Use Cases

[Florin Coras <[e-mail]>]

Hi Ron,

On 10/11/14 4:51 PM, Ronald Bonica wrote:
> Folks,
>
> In Section 2.1, we say that LISP is built on top of four basic design principles:
>
>     - Locator/Identifier split
>     - Overlay architecture
>     - Decoupled data and control-plane
>     - Incremental deployability
>
> However, none of these design principles are unique to LISP. The IETF has produced many overlay architectures over the years and nearly all of them share these characteristics.

I could stand corrected, but out of the ones mentioned in RFC 6115, and 
which have also become RFCs, LISP is the only one with this set of 
properties.
>
> Oddly, the one design principle that *is* truly unique to LISP is omitted from the list. That is, the route pull model.

Nothing precludes the use of a "push model"  mapping system. That is, 
BGP could be used as a mapping-system as long as an adaptation layer as 
per RFC6833 is implemented.

>
> Likewise, In Section 7, we site several use cases to which LISP might be applied. However, we say nothing about why LISP might provide a better solution than any of the other overlay architectures that the IETF has produced in years gone by. Does LISP provide a superior solution because of its one unique characteristic?
>
> In order to fix these problems, I suggest that we make the following changes to Section 2.1:
>
> - add a bullet concerning route pull

Agreed.
> - add a sentence saying that route pull is the only principle that is unique to LISP

With this however, considering the points above, I disagree.

Florin
>
> A use case should be included in Section 7 only if route pulling makes the LISP solution superior to existing solutions.
>
> Ron Bonica
>
> _______________________________________________
> lisp mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/lisp