The Grid: The Good, The Bad, and The Ugly

1
The Grid The Good, The Bad, and The Ugly

• Good
• Simulation and Prediction
• Simulation Tools enable
  Systematic Design
  and Evaluation
• Developing functional software
• Bad
• Point solution to apps
• Still too hard to use
• Reproducibility, scalability, and usability
• Underestimated the technical difficulties
• Not enough of a scientific discipline
• Oversold the grid concept?
• Solution in search of a problem
• Ugly
• Gap between hype and reality
• GGF

2
Where are the 'new' areas in tool development?

• Good to have an Informal Survey of Aps people
• Ask roughly 25 grid system developers,
  application developers, users, etc.

3
Computer Scientists vs. Application Users and the
Grid

• Enjoy complexity because
• Its cool

• Hate complexity because
• It makes life harder

4
Computer Scientists vs. Application Users
How can I solvemy scientificproblem?
How can I make itfaster, better, cooler, ?
Its all about the science
5
Issues

• Grids and Grid middleware are everywhere
• Grid applications are lagging behind, big jump
  from prototypes and demonstrations to real
  production use of Grids.
• Problems
• Interfaces are not simple for scientific
  application developers
• Missing or immature grid services
• Changing environment Fault rich
• Different and evolving interfaces to the grid
• Application developers accept Grid computing
  paradigm only slowly

6
Questions?

• Where should funding be targeted in this area?
• Is there any funds available?
• Perhaps we should prepare a survey for grid
  application users on what they want wrt the
  tools?
• What is the GGF doing?

7
GPW 2004

• Several topics were identified as especially
  important for the next workshop. These include
• How Grid operators make use of performance data
• Performance and infrastructure what are better
  ways to build performance hooks into middleware
• Performance engineering on the grid which
  parameters have been used in anger
• What are the right ways to compare monitoring
  systems
• Where do simulations fit into performance studies
• How should classic data sets be defined