Community building with the European Grid of Solar Observations

1
Community building with theEuropean Grid of
Solar Observations

• Andre Csillaghy
• csillag_at_fh-aargau.ch
• www.egso.org
• University of Applied Sciences Northwestern
  Switzerland

2
An EGSO 5-Minute Tourorhow similar problems
might be with EOEGSO is an IST project(U.K.,
France, Italy, U.S., Switzerland)
3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
EGSO Goals

• Provide a tool to make solar data integration a
  routine operation
• Before Grids, integration is the long term issue
• Data and users are scattered around the world
• Raw data are difficult to move
• Associated with the engineers
• Large Terabytes in a few years
• Information integration is a generic topic, IBM
  estimates that in 2004, 60 of the CS
  professionals work on infromation integration
  issues
• Really a community building
• No proprietary solution assumed
• Everyone must benefit
• The system must provide added value
• Grid is one possible solution

7
GUI
8
EGSO about roles and relations
Consumer
Consumer
Consumer
Broker
Broker
Broker
Provider
Provider
Provider
Provider
AGU Results very general
9
Data Flow Across Roles
Brokers
Provider Y
query
results
10
The Grid Hype?

• Today many things are Grids
• I really do believe in Grids, but which grids?
• Foster et al. give us an answer
• A Grid never gets turned off
• A Grid solve more problems than its individual
  components can solve alone
• A Grid is based on open standards
• Is Oracle Grid a Grid ?
• Is Novartis Grid a Grid?
• But perhaps today a Grid tend to be just a
  better Web Service
• See WSRF
• or a just a standard
• OGSA

11
EGSO as a test platform for (not so) new
technologies

• GUI Jetspeed, JSP, Gridspheres
• Workflows ICENI, Triana, Chimera
• Databases XIndice
• Messaging VOTable, JaVOT, OpenJMS
• Computing services CoSEC
• Communication Globus GT3
• Programming Apache Framework
• Web Tomcat

12
About using new technologies

• Hardware and OS are not (big) problems
• The use of new technologies require a significant
  investition of resources
• Documentation, bugs, missing features
• It is difficult to estimate how long it might
  take to correct or add features as long as the
  new technology is not directly under control
• The nearest the technology, the easiest the
  integration
• Example of GT2 - GT3
• Even within the project
• New technologies slow down significantly the
  development speed
• but it tests the new technology!
• Delays might eventually explode
• Really a problem of interfaces
• Definition of interfaces is absoluteley
  fundamental
• Definition of tests associated with these
  interfaces is absolutely fundamental

13
EGSO as a community builder

• Its all about people!
• Its not community building, its communities
  building
• A user community and a team community
• The user community
• Does not show a particular interest in the
  development
• .. but is really interested in the result
• will use it it makes them spare time .
• will use if it brings new scientific results
• The system must match closely their expectations
• But they dont really know what are their
  expectations
• Many diverging expectations technology must
  solve this
• Seem obvious but in practice its not

14
EGSO as a team community builder

• The team community is mixed with the user
  community
• So they play a major role in the community
  building
• They work much more closely
• Two sides the application and the computer
  science
• Easy software developers like to work together
• More difficult
• People have different working traditions (France,
  Italy, U.K., Switzerland)
• People tend to be irrealistic about the goals at
  the beginning
• Need to involve developers early
• People spend a lot of time in specifications that
  will not be used
• Hardest difference of interests between
  application oriented and computer science
  oriented people
• Application need to work
• Software need to test new technologies
• One end frustrates the other end

15
Priorities to bulid an integration tool

• Difficult to talk about long term
• Everything change so fast
• Make the development public early
• Feedback from the community absolutely
  fundamental
• Get scientific results early
• Build the tool in small steps, start building
  early
• Agile software development methods are useful
• Lots of specifications at the beginning does not
  really help a lot
• You get better user feedback
• Spend time on defining interfaces and tests
• Choose well the development team
• A small team working well together is better than
  a large team
• Involve the development team in decisions to
  estimate the development time