WP3 plenary meeting London, Jan 1718, 2006

1
WP3 plenary meetingLondon, Jan 17-18, 2006

• Status of the Virtual Institute
• Marco Danelutto
• Workpackage leader
• University of Pisa - Italy

2
Contents

• Results of the review meeting
• Status of the Virtual Institute activities
• Outline of the meeting

3
Contents

• Results of the review meeting
• Status of the Virtual Institute activities
• Outline of the meeting

4
Results of the review meeting

• Technical and competent reviewers
• Stefano Campadello Nokia research, Finland
• Alexander Schill Dresden Univ. of Technology,
  Germany
• Michael Hoevenaars ING, Amsterdam, The
  Netherlands
• Erich Schikuta University of Vienna, Austria
• Alfred Geiger T-Systems, Germany
• Quite interactive review process
• WP3 presentation (as of slides sent)
• Objectives (tasks)
• Roadmap (D.PM.01)
• Partners (composition, dynamics)
• Results (sample)
• Integration activities (visits, workshops)
• Extra (projects, conferences, etc.)

5
Main observations (WP3)

• Relationship with the rest of the world
• How do you position? Which kind of interaction?
• Reference/prototype implementation needed
• A need fundings!
• Security !!!
• Industry
• Excellence gt fundings and industry interaction
• Industry interaction must be strengthened

6
Main answers

• We are taking into account previous / concurrent
  experiences
• (also as recommended from SAB 05 in Barcelona)
• Security is (to be) one of the research groups
• Task force vs. horizontal task even in WP3
• Industry
• We should do better
• Reference implementation
• Forthcoming (hopefully)
• Lightweight specification for openness

7
Review next JPA

• JPA presented
• evolution of JPA1
• Continuity research groups (named themes )
• More focused research topics
• Involvement in standardization bodies
• Observations
• Standardization bodies participation NEEDED!
• More critic deliverables, outlining differences
  with the existing
• Fundamental M30 (all partners using GCM or
  distinguishing activities w.r.t. GCM) !!!

8
Personal comments

• Good referee process
• Better than expected !
• Nice interaction with referees and officers
• No major comments to wp3
• But the ones we already know!
• Needed a better coordination between partners
• Review result of part of us
• Needed more activity on common tasks
• Some comments may be avoided

9
Contents

• Results of the review meeting
• Status of the Virtual Institute activities
• Outline of the meeting

10
Participants (dynamics)

• FHG
• left almost immediately (December 2004)
• due to financial / management problems
• UNCL
• left workpackage activities at the end of first
  year
• kind of Microsoft effect
• active participation in roadmap definition and
  in WP3 meetings
• 14 participants left
• ISTI/CNR, IC, INRIA, QUB, WWU Muenster, UCAM,
  UCHILE, UNIPASSAU, UNIPI, UOW, UPC, EIA-FR, VUA,
  VTT

11
Partner and first year activities
12
First year financial issues

• Most AC partners were unable to justify MM with
  hired people
• too small fundings
• too late hiring
• At the moment the underspent money is reported to
  the second year
• Situation must be corrected during the second
  year

13
Meetings

• Charleroi meeting
• first de visu meeting of all partners
• Jan 2004, Pisa meeting
• roadmap outlined
• Barcelona meeting (Jun 2005)
• Fractal and research groups
• Sophia Antipolis, Oct 2005 meeting
• first explicit Task 1.1 session
• D.PM.02 design session

14
Electronic meetings

• Decided in Barcelona
• No followup
• - collect addresses (chat, Skype, iChat,
  Messanger, ) on a private WEB page
• - exploit existing framework
• (EIA-FR Switch facilities tested (3 partners))
• - spare explorative and finalization short
  visits

15
Web Site
16
Web site (2)
17
Short visits

• Quite a large number
• 20 during the first year
• Involving 11 of 14 partners
• Main integration tool
• Need to be enhanced
• Web site with the one page report !
• almost nobody is submitting such info ---gt wiki?
• Inactive partners (!)

18
Fellowships

• Fellow Marcelo Pasin (full WP3 fellowship)
• Subject Object oriented environment for HPC
  applications on the GRID
• First institute Haute Ecole Spécialisée de
  Suisse Occidentale with Pierre Kuonen
• Second institute University of Pisa with Marco
  Danelutto
• Duration 14 months
• Fellow Nikolaos Parlavantzas (Joint
  fellowship WP7 WP3)
• Subject Dynamic software components composition
  in GRID environments
• First institute University of Westminster with
  Vladimir Getov
• Second institute INRIA with Denis Caromel
• Duration 18 months
• Fellow Catalin Dumitrescu (Joint fellowship
  WP3 WP6)
• Subject Scheduling algorithms for higher-order
  components on the GRID
• First institute University of Delft
• Second institute University of Muenster
• Duration 18 months

19
Technical Reports

• Currently
• TR-0001 Optimization Techniques for Implementing
  Parallel Skeletons in Distributed Environments
• TR-0002 Behavior Customization of Parallel
  Components for Grid Application Programming
• TR-0007 Characterization of the performance of
  ASSIST programs
• TR-0014 Parallel program/component adaptivity
  management
• TR-0016 Automatic mapping of ASSIST applications
  using process algebra
• Are from WP3 (5 out of 17)
• All papers submitted (joint ones) should become
  TR !!!

20
Pubblications

• About 20 publications by single partners related
  to topics of the Programming model Virtual
  institute
• About 15 joint publications by partners of the
  Virtual institute
• Startup
• more activity expected while going on
• as effect of achieved integration

21
Publication data base

• Thierrys email last week
• SAB recommedation
• All partners
• must contribute
• Useful tool for
• Integration and
• Advertisement!
• Bibtex and PDF
• needed

22
JPA2 MM
23
JPA2 deliverables

• Some committments here
• - workshop at M30

24
JPA2 milestones
25
JPA2 research groups (1)

• Programming models for the single componentwhich
  are the more suitable programming models that can
  be exploited in the implementation of a single
  component
• Communications (inter and intra
  component)point to point and collective
  communication implementation and optimisation
• Primitive component definitionmechanisms and
  formalisms to be used to define a component
• Hierarchical component compositionall the
  details related to the definition of new
  components out of the composition of existing
  ones
• Advanced programming modelsnew programming
  models that offer the programmer more advanced
  tools and higher level mechanisms to implement
  grid applications on top of the component model

26
JPA2 research groups (2)

• Performance modelsthe possibility to model the
  behaviour of component execution in such a way
  unexpected behavior can be discovered by
  monitoring execution parameters and comparing the
  measured parameters with the theoretically
  predicted ones
• Adaptivityall those techniques that can be used
  to automatically/autonomically adapt component
  behavior to grid changing features (faulty nodes
  and links, varying loads, unpredictable hot
  spots, etc.)
• Component run time supportconcerning the
  features/mechanisms needed to run components on
  grids as well as the (optimized) run time systems
  supporting component execution
• Component semanticsconcerning all the aspects
  related to the definition of formal and informal
  component and component compositions semantics
• Interoperabilitythat is all the problems
  related to the possibility to run component
  programs interacting constructively with non
  component software built according to existing
  grid standards as well as with component software
  developed according to different component models

27
JPA2 research groups

• Security
• not in the JPA (fault !) but we must take it
  into account
• Interaction with other CoreGRID work packages
  aimed at providing the other CoreGRID work
  packages with the results produced within WP3 by
  the research groups listed above and gathering
  requisites/requirements and suggestions from the
  other work packages, in particular from those
  that will eventually use the GCM or implement the
  Grid platforms aimed at supporting GCM. Within
  the M13 M30 period, close interaction is
  expected with the other CoreGRID work packages so
  that both those institutes will consider using
  GCM in their specific activities. WP3 will also
  benefit from the suggestions/requirements coming
  from those work packages while refining/developing
  the GCM.
• o Interaction with national/international
  standard institutes/committees aimed at
  promoting the results achieved within WP3 at
  different levels. We expect that during the M13
  M30 period the partners of WP3 will start
  participating in such institutes/committees
  activities and that partners of CoreGRID WP3 will
  participate in committees designing/defining new
  research frameworks at both national and
  international level.

28
Research groups

• Responsibles needed
• Proposal if no resp. and/or no participants gtgtgtgt
  remove (JPA2?)
• Programming models of single component (EIA-FR)
• Communications (WWU-Muenster)
• Primitive component definition (INRIA/OASISIRIS
  A)
• Hierarchical component definition (INRIA/OASIS)
• Advanced programming models (UPC,UNIPI)
• Performance models (ISTI/CNR)
• Adaptivity (UNIPI,INRIA/IRISA)
• Component run time support (VUA)
• Component semantics (INRIA/OASIS)
• Interoperability (UOW)
• Security (UCHILE)
• Interaction with other VI
• Standards

29
Projects (new ones!)

• One proposal in call 5 (sep05) (no news as of
  today, still waiting)
• GridCOMP
• STREP (INRIA leader, UNIPI, UOW, ISTI/CNR )
• provide reference/prototype implementation of
  GCM as of D.PM.02 on top of existing, open
  source middleware
• DFG research action proposed
• MetaGrid Metadata and Metaprogramming for the
  Grid.
• German joint research action (DFG-Schwerpunktprog
  ramm).
• Funding for 20 researchers, 6 years, 9M EUR,
  granting decision expected in May 2006.
• Interaction with CoreGRID planned, WP3
  initiators
• UNIPASSAU, WWU Muenster, ZIB (plus four
  others).

30
Positioning w.r.t. NGG reports (12)

• Software view (with implications for architecture
  and user view) addressing
• Programming grid through abstractions
• GCM combines parallel distributed programming
  
• practices in a coherent way
• Invisible grid scenario
• GCM implementation encapsulates all grid
  related concerns
• dynamicity heterogeneity handling,
  middleware
• interface, resource discovery management,
• Self- issues
• targeted in the run time support / component
  framework
• Currently going to be addressed
• Relationships/requirements on Network Centric OS
• which features can be moved to OS ?

31
Positioning w.r.t. NGG3 (SOKUs)

• NGG3
• public in this period
• moving from services to SOKUs (Service Oriented
  Knoweldge Utilities)
• Still central point (research issue)
• raising the level of abstraction (programming
  environments)
• Comment from a referee what if components are
  no more trendy?

32
Contents

• Results of the review meeting
• Status of the Virtual Institute activities
• Outline of the meeting

33
Program
34
Program

• Wednesday, January 18th
• 9.00 - 9.30 Status of the Institute Danelutto
• 9.30 - 11.00 GCM D.PM.02 Henrio
• 11.00 - 11.30 Coffe break
• 11.30 - 12.30 Discussion on GCM deliverable All
  participants
• 12.30 - 14.30 Lunch
• 14.30 - 15.00 Roadmap 1.0 (D.PM.01) Danelutto
• 15.00 - 16.00 Roadmap 2.0 general
  discussion All participants
• 16.00 - 16.30 Coffe break
• 16.30 - 18.00 Roadmap 2.0 discussion and
  assessment
• All participants
• Thursday, January 19th
• 9.00 - 13.00 Presentation of partner's current
  activity
• (10-20 min each) All participants

35
(No Transcript)
36
(No Transcript)