Re: [dispatch] [Secdispatch] IETF I-D submission - please review draft-sandowicz-httpbis-httpa2

[Martin Thomson <[e-mail]>]

On Fri, Oct 14, 2022, at 11:13, Dan Wing wrote:
> However, as I skimmed through the rest of the draft, I got the sense 
> there are other motivations for the design.  Those motivations are not 
> clear.  

For others reading along, the ArXiv paper[1] is somewhat clearer about motivations, though I still found that I was still left with no real understanding of the goal of the draft.  What I was looking for was a clear articulation of:

1. What specific problem exists. (problem statement)
2. Who is likely to be afflicted by those problems. (scope)
3. What general approach might be used to address these problems. (architecture and overview)
4. Rationale backing those high-level design choices. (design choices, alternatives)
5. Enough detail of how those design choices might be implemented that I can be convinced the design is feasible and broadly appropriate.

The bulk of protocol detail can be omitted for early proposals like this one, with more focus on the earlier pieces.  Both documents seem to assume a fair bit, jumping past a lot of the motivating pieces.

For instance, the paper says "The client can verify those assurances provided by the server to decide whether they want to trust to run the computing workloads on the non-trusted server or not."  Up until this point, there was no mention of workloads at all.  There seem to be some assumptions being made about the application that this protocol is intended to support that isn't explained adequately.

[1] https://arxiv.org/pdf/2110.07954.pdf