Defining the Grid

1
Defining the Grid

• Fabrizio Gagliardi
• EMEA Director
• Technical Computing
• Microsoft Corporation

2
Grids A Catch-All Term

• Grids mean many different things to many
  different people/companies
• P2P desktop cycle-stealing
• Linked Supercomputer Centers
• Managed virtual distributed clusters
• Internet access to giant, distributed
  repositories
• Virtualization of data center IT resources
• Out-sourcing to utility compute centers
• Sharing resources distributed among different
  administrative domains (Ian Foster)
• For Microsoft, Grids are about Data Management as
  much as Compute Cycles

3
A more personal definition

• Essential to the success of the Grid is the
  existence of a widely distributed community
  sharing common goals, prepared to share resources
  (including human resources and skills) to address
  and solve common problems which require computing
  capacity exceeding the limits of more
  conventional distributed systems (distributed
  clusters, remote access to SC centres, etc) and
  all this in the most dynamic possible way
• Particle physics is a good example which added to
  the above requirements a chronic lack of adequate
  funding and a tradition to take risks and break
  with more established IT solutions

4
The EGEE Grid
www.eu-egee.org
5
Current trends

• We discussed yesterday the issues of processor
  VLSI technology limits and the need for new
  radical changes in CPU architecture
• This will have a cascade effect in the entire
  chain from the processor all the way to the final
  end-user application
• Multi-core processors will require multi-thread
  support on chip, in the O/S, in the compiler and
  in the overall S/W development environment and
  application
• The Grid computing model is by its nature a good
  match with this evolution at a macroscopic level

6
Evolution will leverage IT Industrys Existing
trends and RD

• Digital experimentation
• Collaboration-enhanced Office productivity tools
• Structure experiment data and derived results in
  a manner appropriate for human reading/reasoning
  (as opposed to optimizing for query processing
  and/or storage efficiency)
• Enable collaboration among colleagues
• (Scientific) workflow environments
• Automated orchestration
• Visual scripting
• Provenance

• Parallel applications development
• High-productivity IDEs
• Integrated debugging/profiling/tracing/analysis
• Code designer wizards
• Concurrent programming frameworks
• Platform optimizations
• Dynamic, profile-guided optimization
• New programming abstractions
• Distributed systems issues
• Web Services HPC grids
• Security
• Interoperability
• Scalability
• Dynamic Systems Management
• Self (re)configuration tuning
• Reliability availability
• RDMS data mining
• Ease-of-use
• Advanced indexing query processing

7
Where Grids will be in 5 years? (from SC05 panel)

• Like in the past ES, AI, networking, OS Grid will
  disappear from the hot research (and hype) space
  and become mainstream technology
• Major Grids already work in production
• Major IT vendors will integrate Grid middleware
  in their standard products
• ISPs will offer a wide range of services Grid
  based
• A full mature market will develop for these
  services
• Computing and data resources will become
  commodities on the Internet
• The result will be a tremendous computing and
  data processing power to enable new scientific
  applications and generate opportunities for
  business applications
• A potential leveler for a worldwide science and
  economy
• Digital Divide could be moderated
• and time will tell how wrong we are in our
  predictions now