Re: [T2TRG] Review of draft-hong-t2trg-iot-edge-computing-03

[Xavier de Foy <[e-mail]>]

Hi Thomas, thanks for taking the time to review and for the constructive
comments.

We (co-authors) basically agree with your points and will take them into
account in our next update (we will leave more time for other reviews to
come in, if any), please find additional details below.

On Fri, Apr 17, 2020 at 9:26 AM Thomas Fossati <[e-mail]>
wrote:

> Hi Xavier and co-authors,
>
> I've read your draft and found it to be a good document.  To me it looks
> like a very useful addition to the corpus produced by this RG, so I'd
> recommend adoption.
>
> Below I've left a few annotations in case you find them useful.
>
> High level:
> - Nicely structured document;
> - Some sections are fully developed, other less so, which is totally
>   normal at this stage.  In particular, I'm having a bit of trouble
>   with Section 3.5 content-/scope-wise.  I also found forward reference
>   to Appendix A slightly confusing.  It probably needs a bit of reorg
>   to disentangle the various threads into a clearer picture.
>
>
[xavier] Yes, it seems best to cut section 3.5 from appendix A, frame it
better as a current "state of the art" overview and maybe place it later in
the document (e.g. after the challenges on IoT, and before describing the
functions).


> Low level:
>
- Section 3.2.  "It is predicted that by 2019" needs refreshing :-)
> - Please remove section 2 altogether, 2119 language is not needed here;
> - Section 4.1.  Not just the BE nature of the Internet, it may also be
>   simple and pure speed-of-light limitations at work there;
>
- Section 4.3.  The dual is also true, a cloud back-end might want to
>   have a reading of the device even if it's currently asleep.  In this
>   case the edge GW caches the last reading and makes it available to the
>   cloud;
>
- Section 4.4.  There's also a threat from passive observers on the
>   device-to-cloud path that can do traffic analysis even when the
>   communication is encrypted end-to-end.  So hiding the natural traffic
>   patterns associated with the sensor network can be a further
>   requirement?
>

[xavier] Agreed on your low level comments above. This last comment seems
to be a good additional argument to use EC for IoT.

- Appendix A.  It's a good snapshot but it has an intrinsic "aging"
>   problem.  Maybe better keeping its content in a living document and
>   referencing it from here?
>

[xavier] Agreed, keeping the appendix may not be a good long term solution.
I assume by living document you mean another individual draft?


> Nits:
> - "Cloud computing has been a predominant" =>
>        "Cloud computing has become a predominant" maybe?
> - Not clear what "post-clouds" means in "we will reach the era of
>   post-clouds"; Is this an established term? I'd never heard it before.
>
> cheers!
>
>
Best Regards,
Xavier.