Data Grid Breakout

1
Data Grid Breakout

• Pragma 3
• Fukuoka, Japan

2
Working Group Participants

• Philip Papadopoulos, SDSC
• Mitsuhisa Sato
• Ekasit Kijsipongse, NECTEC
• Piyawut Srichaikul, NECTEC
• Baoping Yan, CNIC/CAS
• Kai Nan, CNIC/CAS
• Weiming Chen, ISOC/CAS
• John Hicks, Indiana University
• Osamu Tatebe, GTRC/AIST
• Atsuko Takefusa, Ochanomizu Univ.
• Arthur Chen, Academia Sinica
• Eric Yen, Academia Sinica

3
Challenges

• Get gFarm deployed uniformly on all available
  resources to provide a uniform data abstraction
• Demonstrate 4 sites of distributed data (gfarm,
  SRB)
• Aggregate ?? TBs available to PRAGMA activities
• 2,4,6 TB?
• Web-services/Grid-services model for data
• Demonstrate web service flow control. NECTEC
  already going in this direction for water
  resource data. (java based)
• SRB available for PRAGMA Members? What is the
  relationship between gFarm and SRB.
• Is there a capability to demonstrate at SC2003?

4
Databases?

• XML databases for meta-data. Xindices is software
  from Italy (open source part of apache). want
  to share experiences
• Schema for SDB (CAS).
• There are lots of schemas in many places. Are
  there PRAGMA applications that need some
  unification?
• CAS is working on a web-services approach to
  providing program-level access to the SDB.
• Lattice QCD datagrid ? Issues
• Replica Mgmt
• Size of Data

5
GFarm

• Identify sites
• Initial SC2002 Sites (IU, SDSC, AIST, Titech)
• Other sites identified by GGF-7
• CAS, Academia Sinica, Univ. Tsukuba?
• Identify a coordinator
• Technical Coordinator is Tatebe-san
• Need a resource coordinator
• Each site needs a point of contact/person to
  learn gFarm
• Have gFarm software deployed at testbed sites by
  April 30th.
• NECTEC to try their application with initial
  report PRAGMA-4 (also can try client-side only),
  (SDSC, too)
• At PRAGMA-4 identify other applications that
  might be able to take advantage of this testbed

6
SRB

• Purpose to compare utility of SRB vs. gFarm for
  applications
• Performance
• Ease of use
• Stability
• Start at/after PRAGMA-4
• By PRAGMA-4, Phil P. to find out the details of
  SRB licensing for PRAGMA.

7
Web Services/Grid Services Model

• Status Report of where NECTEC development is by
  PRAGMA-4
• In Melbourne, clarify plans for the second half
  of the year.

8
Physical Storage Resources

• Closely aligned with gFarm deployment.
• Have an aggregrate of 2TBs by 30 April 2003.
• SDSC already has 250 GB under gFarm, so total
  goal looks reasonable.

9
Plans beyond PRAGMA-4

• Need to visit what could be demonstrated at
  SC2003.
• Interaction with PRAGMA applications groups
• Support bandwidth data challenge again, but with
  a larger number of institutions supporting the
  effort.
• SRB, gFarm comparison
• Again plans to be clarified at PRAGMA-4