LongTerm Benefits of GRID in the Space Domain

1
Long-Term Benefits of GRID in the Space Domain

• C.C. Harvey
• Centre de Données de la Physique des Plasma
• CNRS/CESR, Toulouse, France
• Disclaimer the author is NOT a GRID specialist
  !

2
What is GRID ?

• GRID is a concept which appears to have been
  invented in 1993 by Ian Foster.
• It aims to share computer resources, seamlessly,
  and on a global scale. The desired advantages
  are obvious
• exploitation of resources which are temporarily
  idle
• immediate access to data
• easier interfaces between codes executing on
  different sites
• GRID seems not to be an acronym in English. But
  in French it is the acronym for a near-perfect
  description of GRID  Globalisation des
  Ressources Informatiques et des Données .
• A recent résumé of GRID progress has been given
  by Ian Foster in Physics Today Vol 56, no. 2,
  pp 42-47 (February 2002).

3
What is GRID ?

• Just as
• the Internet capability of exchanging files
  between computers has revolutionised scientific,
  industrial and commercial activity,
• so the GRID capability of sharing resources may
  be expected to revolutionise programming
  techniques and computer performance.
• But
• the Internet had, and still has, its security
  problems,
• so the GRID will also have both anticipated and
  unexpected security problems.
• The expected benefits
• have been amply demonstrated by SETI (Search for
  Extra-Terrestrial Intelligence), but
• in the Space Domain they will probably be much
  longer term.

4
What is GRID ?

• Grid is a system of
• parallel and distributed autonomous systems,
  which enables
• sharing, selection, and aggregation of
• geographically distributed autonomous
  resources
• dynamically at runtime
• depending on their availability, capability,
  performance, cost, and the users
  quality-of-service requirements.
• These resources include
• processing capability,
• storage capacity,
• data, which must be
• independent of its geographical location
• in a form which is immediately usable

5
What is the Space Domain ?

• Everything which concerns space activities. In
  arbitrary order
• Spacecraft engineering
• Spacecraft operations
• Earth resources
• Space weather
• Space science
• For each category, the following questions can be
  asked
• is GRID essential,
• is it merely useful
• how much is one prepared to pay to obtain it, in
  terms of
• additional costs
• acceptable risks

6
Spacecraft Engineering

• Subsystems and experiments, which are fabricated
  at different sites, have to be integrated into
  every spacecraft.
• Then the ensemble undergoes environmental tests.
• By their very nature, integration and testing
  conducted in special facilities, over a
  non-negligible period of time.
• The only occupation for the test-site subsystem
  and experiment technicians is data verification.
  The presence of the system engineer is not
  justified.
• However, any problems detected must be remedied
  rapidly.
• The system engineer in his home institute needs
  full access to test data, and good communications
  with his technicians at the test facility.

7
Earth Resources

• Earth resources requires comparison of data
  obtained from different instruments on different
  satellites.
• Or use of image processing tools on remote sites
• I imagine that these applications may be able to
  use GRID architecture if it proves economic to do
  so.

8
Space Weather

• A plausible approach to predicting space weather
  involves feeding, immediately, recently acquired
  observational data into powerful models, which
  most probably involve numerical simulation.
• GRID seems an absolute necessity for this
  approach
• The space weather community seems highly
  motivated to use GRID.
• Another aspect of space weather is improvement of
  the models. This can be put under the heading
  Space Science

9
Space Science

• Space science is like all other science it
  would like to profit from use of GRID !
• At the SpaceGRID workshop last October we heard
  how useful sharing of computer resources would be
  for research in cosmology.
• We will hear today about the European Grid of
  Solar Observations and the Astrophysical
  Virtual Observatory
• Progress in planetary science could be made if
  codes for atmospheric and exospheric simulation
  (on different machines at different sites) could
  be interfaced and executed in parallel.
• We also heard about the need for database
  interoperability.

10
Database Interoperability (1/3)

• I develop this point because I have some
  first-hand experience of the relevance of GRID.
• I encourage other participants to share GRID
  experiences with us.
• Interoperability is the ability
• to find, to recover, and to use
• data wherever it may be located on the Internet.
• Clearly
• the data required can be neither found nor
  exploited easily unless described in a
  machine-understandable way.
• any tools developed to handle the mechanics of
  data searching and exploitation will be
  short-lived if confined to space plasma physics
• space plasma physics is not the only discipline
  to encounter this information technology problem
  !

11
Database Interoperability (2/3)

• After having submitted an EoI for the EU 6th
  Framework Programme, I discovered that
• 1) The formalism use to make the data description
  machine interpretable is the subject of Ontology,
  which has recently enjoyed a remarkable revival
  after being dormant for some 150 years. (Of
  course, the key words to be fitted into the
  formalism must be provided by space scientists.)
• 2) The use of the data description to decide
  which of multiple answers best meets the users
  description is the subject of Mediation, which is
  also actively being studied.
• 3) That NASA/OSS seems likely to invest in this
  activity for the needs of NASA Space Science.

12
Database Interoperability (3/3)

• Questions
• Could I have discovered this information more
  rapidly via some sort of SpaceGRID activity ?
• Could some sort of SpaceGRID activity be useful
  now ?
• Observations
• I do not think that pulling scientific data off
  the Internet and automatically feeding it (in
  real time) to research applications is something
  which I want to do !
• Nevertheless, this functionality will be required
  for Space Weather but this is an operational
  service using data from known, validated
  partners.
• I do want to get my hand on data, quickly and
  easily.

13
SPASE

• Space Physics Archive Search and Exchange (SPASE)
  is a consortium which results from a resolution
  of an ISTP workshop in RAL in September 1998.
• Until recently five participating data centre
  CDPP, RAL, NSSDC, SwRI, PDS.
• Since its presentation at the AGU meeting last
  December,n interest in USA has been quite
  remarkable
• Interest in Europe (presented at EGU-AGU meeting
  in Nice) is considerably less (because funding is
  less evident ?).
• One of the things which we urgently need is
  cross-discipline user scenarios for example
• space-physics/solar physics, or
• space-physics/atmospheric physics
• Could SpaceGRID help ?

14
Summary of the Situation

• Numerous communities in the Space Domain are
  interested in using GRID facilities, to greater
  or less extent depending upon the cost and the
  advantages to them.
• Furthermore, they have specific requirements
  which may or may not be being addressed
  elsewhere.
• GRID development in the Space Domain (as
  elsewhere) requires a unique blend of scientific
  and technical competence.
• Sometimes the borderline between technical and
  political questions is also difficult to define,
  for example .....

15
The European Context

• One particularity of the Space Domain is the
  multiplicity of funding
• European funding (ESA, EU)
• national funding (which is itself sometimes the
  responsibility of more than one agency).
• Some GRID activities, such as coordination of
  archiving activities (who archives what, and
  where) may be able to profit from a European
  forum for discussion.
• Who in Europe is responsible for Outreach and
  Educational programmes ?
• These is more  political  questions.

16
Questions

• 1) To what extent are these specific requirements
  being addressed elsewhere (e.g., W3C, ...) ?
• 2) If so, how do we in the Space Domain gain
  access to the most recent information ?
• 3) What is the specificity, if any, of
  applications of GRID architecture in the Space
  Domain ?
• 4) Is there a need for some coordination of GRID
  activity within the Space Domain ?
• The answers to these questions are not obvious
  otherwise, we would not be still asking
  them today !

17
Specificity of Space Domain applications of
GRID

• 1) The applications mentioned all concern
  projects related in some way to ESA.
• Is this a sufficient argument for a strong ESA
  engagement ?
• 2) Some applications (s/c integration, space
  weather, ... ) may interact directly with ESA
  infra-structure.
• A much stronger argument for ESA to become
  involved.
• 3) Some GRID activities (data description for
  space science, ... ) require specific competence
  from the space domain.
• Note that the development of GRID tools should be
  inter-disciplinary, but their application is not.

18
Possible Benefits of SpaceGRID activity

• 1) Enable people confronted with problems to seek
  advice within a familiar scientific community
• 2) Help establish a line of communication with
  (or redirect questions towards) the international
  organisations which deal with GRID problems
• 3) Ensure that on-going work does not become
  needlessly incompatible with GRID architecture
• 4) Exchange ideas and maximise compatibility
  between neighbouring space science disciplines
  (e.g., solar physics, space physics, atmospheric
  physics).
• 5) Encourage an interest in GRID in areas where
  it does not yet exist, and could be useful.

19
Conclusion (1/2)

• Many user communities hope and expect that GRID
  will bring substantial benefits.
• GRID is evolving quickly, but it is still
  immature.
• While waiting, these communities must continue to
  develop their own applications, while remaining
  mutually coherent.
• This is important, both
• intrinsically, and
• with respect to our funding agencies (credibility
  of our proposals).
• Furthermore, support may be found from sources
  other than scientific or space research
• the information society
• education and training
• outreach
• provided that our different work plans are
  mutually coherent.

20
Conclusion (2/2)

• The Space Domain is not going to develop the
  GRID,
• but it wants the functionality promised by GRID
• Can SpaceGRID
• help users to help themselves
• for the overall benefit of the Space Domain ?
• If so, how should it be organised ?
• This is what this workshop must try to determine.

21
End of the presentation Thanks
22
Useful references

• Grid Computing Info Centre,
  http//www.gridcomputing.com/
• SETI_at_home the Search for ExtraTerrestrial
  Intelligence http//setiathome.ssl.berkeley.edu/