Ruminations%20on%20Grids

1
Ruminations on Grids

• Paul Messina
• Center for Advanced Computing Research
• California Institute of Technology
• February 18, 2002

2
GGF External Advisory Committee (GFAC)

• Current GFAC
• Frederica Darema (US NSF)
• Bill Feiereisen (LANL)
• Fabrizio Gagliardi (CERN)
• John Hurley (CAU)
• Paul Messina (Caltech)chair
• Yoichi Muraoka (Waseda Univ)
• Alexander Reinefeld (ZIB)
• Rick Stevens (ANL)

• Proposed New Members
• Irving Wladawsky-Berger (IBM)
• Tony Hey (UK)
• Mary Anne Scott (US DOE)
• Sangsan Lee (KISTI)
• Kyriakos Baxevanidis (EU/CEC)

3
GFAC Role

• GFAC provides external advice on strategy to the
  GGF chair and GFSG
• GFAC annually reviews the performance of the GGF
  chair and approves appointment or reappointment
  of the GGF chair every 3 years
• GFSG nominates chair candidates

4
Issues GFAC deals with

• Community and regional representation in GFAC and
  GFSG
• Size of leadership team (GFSG)
• Making the GGF standards process a success
• e.g., ways of incentivizing writing detailed
  proposals and position papers
• Relationships with other activities such as
  Peer-to-Peer Working Group and New Productivity
  initiative

5
P2P

• Peer-to-peer computing is the sharing of computer
  resources and services by direct exchange between
  systems
• These resources and services include the exchange
  of information, processing cycles, cache storage,
  and disk storage for files
• The Peer-to-Peer Working Group was organized to
  facilitate and accelerate the advancement of
  infrastructure best-known practices for
  peer-to-peer computing

6
NPi

• NPis primary goal is to specify a set of
  standards for effective interoperability across
  the Distributed Resource Management (DRM) space
• With DRM standards in place it will be easier to
  build efficient, multi-vendor distributed
  computing solutions
• NPi is an open, industry-wide group of leading
  system vendors, independent software vendors and
  service providers

7
Web services or Grid middleware?

• Isnt it both, because each provides something?
• When a technology has mass markets, its a good
  idea to use it where possible
• so we should explore the use of Web services in
  the context of Grids and GGF activities
• some Grid application projects provide a testing
  ground
• Which one wins?
• wrong question success is that eventually there
  is no distinction between the two

8
Lessons from the past

• As my gray beard indicates, I am old enough to
  have been involved in Grid-like stuff for some
  time
• CASA
• I-Way
• SF Express
• The Grid bandwagon
• my evangelism fed back to me

9
The Grid and the Web

• As the Web became an everyday tool for
  everyone, the Grid concept became easy to
  explain
• Web users became familiar with accessing remote
  resources
• typically the resources accessed are static
  documents but some are dynamic.
• Consequently, the idea of harnessing major remote
  computational and data resources was no longer
  quite so foreign.

10
The Grid and the Web (2)

• Second, most institutions installed higher speed
  connections to the internet as demand increased
  and prices fell
• Third, researchers began to put more and more
  data collections on line and accessible to
  others,
• facilitated by the additional trend of rapidly
  decreasing data storage costs

11
Some other standards activities

• Ada
• a standard without experience to back it
• F77
• not much new beyond F66
• In 1974, X3J3 said it was already too late for
  new proposals to be considered
• Stu Feldmans f77 proved to be quite useful!

12
Some other standards activities (2)

• MPI became a de facto standard at least ten years
  after initial attempts to standardize
  message-passing libraries
• once the community was ready, process took a
  small number of years
• compare with F77
• although de facto, MPI has been widely adopted
  and deployed by industry
• with MPICH serving as the foundation of most
  initial vendor implementations
• is GGF ten years too early?
• nahh

13
So?

• Standardization needs to come at the right time
  in the life cycle of a new technology
• Reference implementations are extremely important
  and useful

14
Insomnia

• The Grid is a premier example of
  applications-driven RD inspired by the
  confluence of several technological trends
• applications push and technology pull, great
  stuff
• But applications projects starting to say they
  have a job to do, its too late to take a
  long-range approach, they must focus on solving
  their problems by whatever local optimizations
  they can devise

15
Insomnia (2)

• Many, many Grid projects world-wide
• this is good
• Very high expectations of quick success
• this is scary
• difficult to erase negative initial impressions

16
Summary

• We have many challenges as we try to craft
  standards at the same time as operational,
  production Grids are being implemented
• and the general computing environment is evolving
  rapidly
• Somehow we need to strike the right balance
  between elegance and timeliness, between
  innovation and exploiting existing technologies
• Writing and critiquing proposals is important, as
  is developing reference implementations

17
Summary

• Please contact any and all GFAC members and help
  us help GGF

18
GGF Town Meeting- Wednesday

• GGF Leadership and Structure
• Managing Growth
• Membership Program
• Toward A Grid Professional Society
• GGF, Incorporated
• Current and Planned GGF Support Infrastructure
• Plans for GGF Future Venues and Events
• Beyond GGF6