Re: [decade] reorganized -req

["Y. Richard Yang" <[e-mail]>]

Hi Peng,

Thanks a lot for the feedback! Please see below.

On Tuesday, July 10, 2012, Zhang Peng wrote:

> Hi Richard and All,
>
>         Here are some comments on the recently revised version of -req
> document.
>
>         First, I noticed that the protocol flow including DRP and SDT is
> included in -req. I wonder if this is a good choice, since it is more like
> a design rather than a requirement. Could someone explain the rationale?
>
> I feel that they represent a group of functions (and hence represent two
functional components). They help to organize the conceptual model. The
requirements do not use SDT and DRP. What do you think?

        For the naming requirements, I suggest that the "the naming scheme
> should support that consumer can validate the name-object binding, i.e.,
> can check that the object content is what the name claims to be". The
> rationale is that consumer now obtains the object from DECADE server, and
> in turn is uploaded by some content provider that is not trustworthy.
>
> Sounds good.


>         I suggest that the terminology 2.5 "Provider" should better be
> "content provider". Otherwise, it may be confused with "DECADE storage
> provider" of 2.3. Also, 2.9 is referring to "content provider", not
> "provider". Correspondingly, "consumer" should be "content consumer". But
> we should also comment that we would use shorter term "provider" and
> "consumer" if there is no ambiguity.
>
> I thought about this. It is not content provider but rather resource
provider, because A could allow B to upload to A's account, where A owns
the account resource. I am fine to go with Resource Provider though.

        Section 7.2 is not so clear to me. The rationale has not stated the
> reason for this very well.


I will take a look and get back soon.


>
>         In section 8.1, the term "user" is not defined by us. Should we
> use "client" instead?


Sounds good.

>
>         In Section 9, should we also include "overload condition" which
> appear in section 4.1.3.1 of current version?


I feel that overload is a type of insufficient resources. One possibility
is to distinguish between system overload and insufficient inividual
provider resources. What do you think? But this can be tricky in that the
system can be oversubscribed. So the error message may need to indicate
such.


>         Some minor revisions:
>
>                 - Section 2.3: "that" should be removed;
>                 - Section 6.1:  Should be "After a DECADE-compatible
> client owns an account on a DECADE-compatible server, it should be able to
> read data from and write data to the server once authorized by the
> server.", and it should be following section 6.2.
>                 - Section 6.2: I think it is better to be "Access to an
> account on a server MUST be granted to a provider based on cryptographic
> security"
>                 - Section 6.3 "resoutces" -> "resources"
>                 - Section 6.9 "disovery" -> "discovery"
>                 - Section 7.5 "enerate" -> "generate"
>                 - Section 8.4 "stoarge" -> "storage"
>
>
> It will be great if you can take a pass. Thanks a bunch!

Richard

>
>         I would like to help revise if you don't have time.
>
> B,
>
> Peng.
>
>
> On Jul 7, 2012, at 12:46 PM, Y. Richard Yang wrote:
>
> > Dear all,
> >
> > We did a substantial reorganization of the -req documents. It is
> attached to this email. Any early feedback will be greatly appreciated.
> >
> > Thanks!
> >
> > Richard
> > <draft-ietf-decade-reqs-all_v06.txt><draft-ietf-decade-reqs-all_v06.xml>
>
>