[Nfvrg] Review of draft-natarajan-nfvrg-containers-for-nfv

[Kostas Pentikousis <[e-mail]>]

Hi Diego, all,

| The current draft is available at:
| 
| https://datatracker.ietf.org/doc/draft-natarajan-nfvrg-containers-for-nfv/

It seems that the draft expired at the time of this review. I had a cursory look and I have a few comments on the expired version -03.

First, it would be good to be more explicit about the scope, goal and applicability of the draft. 

Second, the draft does not cover salient research papers in this very active area. In fact, the first reference is to a wikipedia article; only one peer-reviewed paper is cited; most references are web pages. This should be significantly improved; see also RFC 5743, sec. 2.1.

Up to section 4, we only read text with high-level challenges (e.g. "When a resource component is compromised, quarantine the compromised entity but ensure service continuity for other resources") or requirements (e.g. "a management solution ... Is secure"). The conclusion explicitly says that the draft "presented the challenges when building an NFV platform", thus I'm not sure why the draft title includes the word "analysis".

Section 4 reports on a particular set of results. Was there any consultation with bmwg, for instance? I think I mentioned it in the last meeting in Berlin that some more information regarding the experiments is needed. It's not clear how many samples do you have, or what do the reported numbers represent (first run, mean, median, ...), what is the impact of the platform and the respective components, or what new insights we get from this particular set of experiments compared to other work. In general, I think sec. 4 is more suitable for a peer-reviewed workshop paper to be cited by the draft.

Finally, the conclusion is not particularly enlightening: "We conclude that choosing a solution is nuanced, and depends on how much value different NFV operators place on criteria such as strong isolation, performance and compatibility with applications and management frameworks." I would appreciate something more specific.

A few nits as well: On the header (p. 2 ff.), the draft is dated "September 2015", not as on the first page (July 8, 2016). Indentation on 3rd level headings looks strange, ditto for bullet style (cf. lists in sec. 3.1.1 vs. 3.2.1, and the remainder of the document). I'm not sure why there's a need for a single 3rd level section in both cases; ditto for 3.3.1 and 3.4.1.

To sum up, I think more work is needed prior to RG adoption.

Best regards,

Kostas