This is a static archive of the previous Open Grid Forum Redmine content management system saved from host redmine.ogf.org file /dmsf_files/7355?download= at Fri, 04 Nov 2022 20:06:40 GMT Notes of the meeting of the Applications Research Group (APPS-RG),

Notes of the meeting of the Applications Research Group (APPS-RG),

held at GGF9 in Chicago, October 6, 2003

 

 

Notes taken by: Shantenu Jha, <s.jha@ucl.ac.uk>

 

Session chaired by: Thomas Hinke

 

Introduction:

 

Reviewed goals of APPS-RG, as discussed at [Oct 2002, GGF6]

 

1. Goal: Apps Workshop

Reviewed accomplishments e.g. GGF5 in Glasgow, GGF7 in Tokyo andGGF8 in Seattle.

 

One thought- In general, summer GGF for workshops

 

Simon Cox: Should we investigate possiblity of holding WS inFrankfurt with GGF10? (rationale similar to Tokyo).

 

Action Item: Simon Cox to contact Mark Parsons for one lastattempt at organizing a WS at GGF10 (Frankfurt).

 

 

Ray Bear: Challenges of Large Applications in DistributedEnvironments [CLADE].   Lastyear was published seperate to HPDC (held in Seattle). Hinke: Do we needanother WS if CLADE is already happening in the time frame of Honolulu?

 

Rob Hood: Mentioned UPDT will be having a WS at GGF11.

 

Cox: If jointly done with CLADE, usual registration relatedproblems.

 

Hinke: 1 workshop a year, logistics- offsite? venue?

Cox: Possibility of any off-track events to be held at NeSC(Edinburgh,UK)

Dave Berry seconded it.

 

Hinke: Closing remarks: needs further thought.

 

2. Goal: Grid technology assessment

"Build a set of grid applications to evaluate the currentstate of the grid and document the results. Collect statistics of current gridstatus with links to different grids... "

 

Demo-based Grid Assessment Issues:

 

Need clear objective and assoc. metric for demo such as one ormore of the following:

 

 e.g. Ability toeasily use grids

á  abilityof grid to support use of large number of processors across many differentgrids

á  abilityof grids to support high data rate applications

     ???

  Desirable:

á  usedemo to accomplish something useful

á  involvestudent teams

á  involvedthe computational resource on the floor of SC03

    ....

 

Discussion:

 

Where do we want to go with this?

Volunteers to lead and contribute?

 

   ** Don't do anassessment **

 

 

3. Case Studies

Collect case studies of grid applications with a focus on what theapplication does and how it uses the grid to accomplish its goals.

 

Collect 3 types of case studies:

á  productionuse of the grid

á  developmentuse of the grid

á  scenariosof what users would like to do, but can not currently

á  do withthe grid

 

Make recommendations that go back to the other GGFresearch/working groups

 

Cox: make the case studies as a basis for the workshop; anddocument derived from individual contributions for the workshop.

 

classification: compute intensive, data intensive...

 

Franz:  contributeenvironment risk analysis

MPI code, weather

Peggy: industrial application (engineering) used at EFBL, France.

 

Kum Won Cho : CFD

             [possible presentation at workshop,documentation and even live

             demonstration]

 

 ** consensus: explorepossibility of having this in Frankfurt. **

 

Dave Barry: mentioned OGSA use case document. Explore overlap?

Action: Dave Barry to dig out the template of the OGSA Use caseand email to some people...

 

 

4. Goal: Identitify and assess burning issues that applicationdevelopers have that could either facilitate or hinder the development of gridapplications

 

Hinke: GT3 problems (Astronomy BoF)

Susan: insufficient documentation

 

5. Goal: Application-unique requirements --  capture application-unique requirementsfor the grid and make recommendations to other GGF research/working groups.

 

discussion: mailing list.

what do we expect to come out of the mailing list?

 

6. Goal: Identify and classify application-orientedcapabilities that grids currently facilitate

 

Cox: to be used in an executive summary?

 

 

 

Hinke: What else would people have liked to see?

 

á  GridLabrepresentative expressed surprise at the little interest?

á  convincingscientist(s) to use the grid is difficult.

á  usingthe grid is viewed as a problem. Use case will be useful.

á  Nosense of Best practises?

o      griddevelopers guide?

 

concentrate not so much on the application but focus on the gridtechnology e.g. used globus run, grid-ftp.

 

 Hinke: quickpresentations not necessarily in a workshop context

 

Remarks: Best practises possible only when 'hardened architecture'

Hinke: not necessarily 'best practises', but 'a practise'?

e.g. which grid commands

 

Remarks: akin to design patterns, which are best practises fromcommonly recurring problems

 

e.g. recurring problem: job submission (JS)

possible solution:  JS service factory

 

   ** Use APPS-Gas a forum for discussion between end users and tool

   developers.Structure meetings like a show and tell?**

 

 Piotr  Bala: This RG develop a  testbed (resources, easy way to usecertificates, some  toy  applications  etc.) so that a wider  user community couldcome test it, use it and decide if it is for them. Something like this existsfor UNICORE.

 

 

This is a static archive of the previous Open Grid Forum Redmine content management system saved from host redmine.ogf.org file /dmsf_files/7355?download= at Fri, 04 Nov 2022 20:06:41 GMT