Title: Outcome of the EU review of the DataGRID Project
1Outcome of the EU review of theDataGRID Project
- Peter Kunszt
- WP2 Manager
- Peter.Kunszt_at_cern.ch
2Outline
- Review Agenda
- The Result
- The Review
- Introduction and Achievements
- First year objectives
- Architecture overview, issues and actions
- Job submission example (DEMO)
- Plans for 2002 and Issues
- Summary
- The Reactions
- Paris Conference
3Review Agenda Friday 1st March
- Introduction Kyriakos Baxevanidis/EU Hans
Hoffman/CERN - Project overview Fabrizio Gagliardi
- Architecture, middleware delivery schedule
Bob Jones - Applications
- WP8 Federico Carminati
- WP9 Julian Linford
- WP10 Johan Montagnat
- Testbed 1 Charles Loomis
- Demos
- WP8 Eric Van Herwijnen Joel Closier
- WP9 Johan Van De Vegte Christine Leroy
- WP10 Vincent Breton Yannick Legre
- Quality Assurance Gabriel Zaquine
- Dissemination Roberto Puccinelli
4The Result
5Review Feedback Positive Comments
- Comments by Dr Karsten Decker
- Project is congratulated for exceeding
expectations and 1st year objectives it has
been well managed and is on track - The project management and the project office
were especially congratulated for an excellent
job - All deliverables have been delivered on time
before the review and they are accepted as is a
1st for the reviewers - Reviewers welcome incremental releases this
will improve testing and feedback - Demo was useful and realistic not everything
worked to perfection
6Review Feedback Constructive Critique
- Must publish results widely in conferences and
journals - ensure EDG innovations are correctly
credited to the project and EU - Licensing issues are important - urged to solve
this quickly - Data mgmt should be extended beyond simple read
facilities to make it attractive to other user
groups - In future reviews will look for details of
scalability, fault tolerance security - Dissemination needs quantitative results for next
review e.g. number of contacts made at each
event - Some partners overspending need to be understood
in detail.
7Project Review - actions
- Extra-ordinary review to be held in May/June at
ESA site written details of funding activity
requested - Written report from reviewers will be produced in
3 weeks - Kyriakos Baxevanidis congratulations to the
projectplease take on-board the recommendations
from the reviewers
8The Review
9Introduction
- Difficult to review a project of this size and
complexity in one day - Assume all details in the deliverables and
periodic reports - Decision to present the project from the
applications angle - Coherent with the original mission of
demonstrator and test bed for high performance
research networks (RN Geant) - Innovative middleware and fabric developments
presented in a practical demo
10Achievements of the project
- The project is up and running!
- All 21 partners are now contributing at
contractual level (took much time and effort) - Project administrative and managerial structure
established with minimum resources - Important survey of existing technology, market
survey and state-of-the-art performed by most WPs - Globus selected as basis for the middleware
(excellent relations established with Globus
developers in US, stress test of new Globus
releases) - First architecture defined, focused on test bed 1
- Necessary middleware developed by WP1-5, adopted
for future LHC computing project - Packaging, integration and deployment WP6 with
essential contribution by application WPs (8-10)
and support by network and security WP7 - User requirements, test use cases and validation
provided by application WPs (8-10) Ready to
deploy real production for D0 and Babar PP
experiments - Important dissemination activity and quality
control provided by WP11 - Overall management and administration deployed by
WP12
11Objectives for first year of project
- WP1
- Job resource specification, description
scheduling - WP2
- Data access, migration replication
- WP3
- Monitoring infrastructure, directory services
presentation tools - WP4
- Framework for fabric configuration management
automatic sw installation - WP5
- Common interface for Mass Storage Systems
- WP7
- Network services and monitoring
- Collect requirements for middleware
- Take into account requirements from application
groups - Survey current technology
- For all middleware
- Core Services testbed
- Testbed 0 Globus (no middleware)
- First Grid testbed release
- Testbed 1 first release of middleware
12DataGrid Architecture
13Architecture Issues and Actions
- Some concepts remain vague
- e.g. what do we really mean by interactive jobs?
- Some boundaries are unclear
- Between components and between work-packages
- e.g. are we certain about the scope/functionality
of a Storage Element? - Some requirements are not yet addressed
- e.g. anonymous users
- The various software components are not yet fully
integrated - e.g. do the Storage Elements, Computing Elements
Information System work in unison? - Short term/ Long term trade-offs
- Need to satisfy the short-term needs of the
testbed without jeopardizing future directions - e.g. do we patch code now that we know will be
replaced in the next release?
- The reformed architecture group will address
these points taking into account our experience
from testbed1 and further requirements - Implementation of iterative releases and
separation of development testbed from production
testbed - Better sw testing and nightly integration
14A Job Submission Example
Replica Catalogue
Information Service
Resource Broker
Storage Element
Logging Book-keeping
Job Submission Service
Compute Element
15Summary
- Application groups requirements defined and
analysed - Excellent working relationships established with
the application groups (real users) which are
driving forward the overall architecture and
quality of the software produced - Extensive survey of relevant technologies
completed and used as a basis for EDG
developments - Comprehensive architecture defined as a basis for
testbed 1 - First release of the testbed successfully
deployed - Path for extensions/improvements defined taking
into account feedback from previous releases and
needs of the application groups - Excellent collaborative environment developed
with key players in Grid arena - Globus EDG is influencing and participating to
the extensions of the toolkit to ensure it
addresses the needs of the application groups - Other Grid projects Cross-fertilization of
software components with other Grid projects
(PPDG/GriPhyN) - Project can be judged by
- level of "buy-in" by the application groups
- wide-spread usage of EDG software
- number and quality of EDG sw releases
- positive influence on developments of GGF
standards Globus toolkit
16Issues
- The project is under-funded relative to the size,
goals and number of partners - New technology, 21 partners and diversity of
applications and communities (CS, PP, EO and Bio) - Original funding assumptions of CERN (and some
other partners) no longer valid - Project office under staffed and below minimum
operating budget one project secretary, one
technical coordinator and one quality engineer
marginally funded (0 travel budget and personnel
cost higher than expected) - Overall the project within budget, but ½ of
partners overspent travel and 2 largely overspent
first year budget - Almost completely unfunded resources for test bed
equipment, networking and security
17Plans (Technical) for 2002
- Extension of testbed
- more users, sites nodes-per-site
- Iterative releases up to testbed 2
- incrementally extend functionality provided via
each Work Package - better integrate the components
- improve stability
- Testbed 2 (fall 2002)
- Anticipate impact of OGSA
- Ian Foster Carl Kesselman members of reformed
architecture group - Ensure convergence with US Grid project
activities (PPDG/GriPhyN) - to be addressed via InterGrid DataTAG activities
- Planned intermediate release schedule
- TestBed 1 November 2001
- Release 1.1 January 2002
- Release 1.2 March 2002
- Release 1.3 May 2002
- Release 1.4 July 2002
- TestBed 2 September 2002
- Similar schedule will be organised for 2003
- Each release includes
- feedback from use of previous release by
application groups - planned improvements/extension by middle-ware WPs
- use of software infrastructure
- feeds into architecture group
18Future Plans (Administrative/Financial)
- Restructure coordination budget (convert
overheads in travel, convert part of IR Forum in
Project Office administrative staff) - Merge project management and project office with
DataTAG and integrate dissemination and technical
coordination with GridSTART - Economy of scale, synergy of activity
- Explore new unfunded resources PP LCG, national
Grids (GridPP, INFN Grid, French research Grid,
etc.) - Prepare larger follow-up proposal for FP6
19The Reactions
20Review Feedback First Reactions
- Must publish results widely in conferences and
journals - ensure EDG innovations are correctly
credited to the project and EU - Licensing issues are important - urged to solve
this quickly - Data mgmt should be extended beyond simple read
facilities to make it attractive to other user
groups - In future reviews will look for details of
scalability, fault tolerance security - Dissemination needs quantitative results for next
review e.g. number of contacts made at each
event - Some partners overspending need to be understood
in detail.
Papers in preparation for HPDC, GGF, SC2002,
iGrid, grid2002
Agreed on BSD-like licensing model in Paris.
The update mechanisms are in the design phase and
on track.
Security Group and Architecture Group will deal
with these Issues in great detail
Mechanisms are being set up
Review of ESA in June
21Paris DataGrid Conference
- Public Industry Research Forum organised by the
French Ministry of Research high visibility and
publicity - WP coordination and planning sessions in parallel
- Interoperability
- Reactions to review
- Plenary reports
- Status of each WP
- Architecture Group
- Security Group
- Other Grid Projects and relationships