[ghpn-wg] update on the network services use cases draft

[Franco Travostino]

Thank you, Tiziana. I like the approach of splitting the material
between the 1st Tier cases (the most mature ones) and the 2nd Tier
developing ones.

Hereafter my comments:

** Abstract and Introduction have the words "high-level but formal
description". I hesitate to use the word formal (which could lead to
believe that we use UML2 language or something). I would qualify it as
"high-level, structured description" ... that's as far as we've gone
IMO.

** I believe that the title "High-Throughput File Transport with
Deadline" does not do justice to the text in 2.2. I propose to rename
it "Large Data  Streaming coordinated with Job Execution". The
rationale for this change is that the former title does NOT justify
<start quote>"the existence of a Guaranteed Data Rate Service assuring
the requested bandwidth between two endpoints to its client by means
of traffic differentiation techniques ... note that this service needs
to be seamlessly available end-to-end.<end quote>. A deadline does not
necessarily imply "a darn imminent deadline". Depending on how close
the deadline is, one could use a delay-tolerant technique such as L7
store-and-forward without any end-to-end requirement and with multiple
hops to (say) exploit large pipes available at slightly different
times ... as long as the final deadline it's being met, and I get
notification of this deadline either being met or failed (By way of
analogy, real-time engineers go to great lengths to distinguish
between real-time and real-fast requirements ... 2.2 talks about
real-fast while its title hints real-time, to me at least). As well,
my proposed title wants to emphasize the coordination with job
execution, which I thought to be a distinguishing aspect of this use
case.

** In 2.3.1, the SE acronym is introduced without explanation. The
paragraphs featuring the Data Scheduler architectture appear abruptly
without any background context and no understanding of how they might
fit with the subsequent HEP Use Cases (also, the two 1-2-3 numbered
lists back to back are confusing). I propose to remove the text
starting with "These scenarios involve different types of file
transfer, specifically" until the end of 2.3.1, and consider it for
insertion after 2.3.5 ... though we should make sure first that it
passes the Use Case eligibility test (it's all about the WHAT rather
than the HOW).

** Prior to enter public comment, we need to expand the set of
so-called path-oriented use cases. At GlobusWorld 05, I was awestruck
by the work that Rolls Royce and UK researchers are  doing to harvest
sensor data from in-flight jet engines and to correlate such data for
early diagnosis of malfunctions. This would make for a superb use case
with mixed wireline, wireless, mobility, and intermittent connectivity
scenarios. I will approach these researchers ... with some luck they
are already part of the GHPN community and will read these lines of
mine.

** In 3.1.4, the CE and SE acronyms need to be expanded (GGF final
documents have a very diverse audience, we cannot assume prior
knowledge).

** With regard to editorial nits ... The font type/size and the lack
of line breaks between paragraphs had my eyes cross over and lines
blur several times. I also recommend page breaks before level 1
Headings (possibly even before level 2 Headings). Just too much ink on
a single sheet of paper!

** Lastly, I'd like to preserve the association between the Use Cases
and the domain experts who provided input. Rather than enumerating
authors after Section 5, I recommend writing "Input provided by ..."
next to either level 1 or level 2 headings. This reflects the high
degree of specialization that these use cases bring to bear.

that's all for now
-franco



On Thu, 24 Feb 2005 17:48:44 +0100, Tiziana Ferrari
<tiziana.ferrari at cnaf.infn.it> wrote:
> Dear all,
> an update on the status of the "Grid network services use cases" draft.
> 
> I have done some final editing work, reformatted the text according to the
> standard, added some missing information.
> With regards to the list of use cases, the latest version of the draft only
> contains the use cases that reached a mature state and went through a number
> of revision cycles. The other good ones are now part of a companion draft.
> 
> The purpose is to push the mature use cases forward and possibly put the
> draft in public comment soon.
> 
> You can find the latest version of the draft and the companion document
> (draft part II) under the GHPN "current drafts" page.
> 
> Regards,
> Tiziana
> 
> ********************************************
> Tiziana Ferrari            Tiziana.Ferrari at cnaf.infn.it
> Italian National Inst. for Nuclear Physics / CNAF
> 
> v.le Berti Pichat 6/2, 40127 BOLOGNA, ITALY
> tel: +39.051.6092.759      fax:+39.051.6092.746
> http://www.cnaf.infn.it/~ferrari
> ********************************************
> 
>