Development in Grid Activities in Singapore

1
Development in Grid Activities in Singapore

• CHEOK Beng Teck Ph.D.
• (cheok_at_ngp.org.sg)
• Director, National Grid Project Office
• Singapore
• www.ngp.org.sg

2
Outline

• Grid Activities in
• Nanyang Technological University
• National University of Singapore
• Institute of High Performance Computing
• Genome Institute of Singapore
• BII - Bio Medical Grid (not covered in this
  presentation)
• National Grid Project (NGP)

3
Nanyang Technological University

• Research Focus
• Middleware development (with IHPC)
• Applications development
• Projects
• Campus Grid testbed
• MEG Data Visualisation (CyberMedia Center, Osaka
  Univ.)
• Integration of SIBBS into Globus
• Meta-scheduling, inter-operability, ...
• Collaborators
• IHPC, BMG, ApGrid
• HPC Resources
• PC Cluster with Cluster software, SingAREN
• AP Science Technology Centre (with Sun
  Microsystems)

Courtesy of A/P Francis Lee, NTU
4
Nanyang Technological University
MEG Data Analysis

• MEG Non-intrusive method for capturing the
  electrical activity in the brain.
• Can help the practitioner to diagnose certain
  brain illness
• Data transfer, analysis diagnosis are applied
  in the computing grid
• Collaboration CyberMedia Center _at_ Osaka
  University (Japan)

Courtesy of A/P Francis Lee, NTU
5
Nanyang Technological University
Protein Alignment (Internal NTU Campus Grid
Project)
Discovered new protein sequence is usually
analyzed by comparison with genetic sequence
databases, which are growing exponentially. To
speedup the analysis, a sequence database is
partitioned the tasks of sequence alignment are
distributed over a computational grid.
Courtesy of A/P Francis Lee, NTU
6
Simple Inter-Domain Bandwidth Broker Signaling
protocol
Nanyang Technological University

• SIBBS
• An open protocol proposed by Internet2 community
• Might be incorporated in the QBone architecture
• A network-to-network interface (NNI) protocol
  defined for communication between inter-domain
  bandwidth broker (BB)

Courtesy of A/P Francis Lee, NTU
7
University of Singapore

• Research Focus
• Middleware Grid Programming
• Projects
• Campus Grid Testbed
• ALICE (a Java-based Lightweight Grid)
• Collaborators
• Centre for Remote Imaging, Sensing Processing
  (CRISP), BII, IHPC, Nanyang Polytechnic (School
  of Life Sciences)
• Start-up Atsuma Technology (www.atsuma.com)
• HPC Resources
• HP Alpha servers (GS320 and ES40), SGI server
  (Origin2000), Linux clusters, SGI/HP/Sun
  workstations

Courtesy of A/P Teo Yong Meng, NUS
8
NUS Grid Testbeds

• Linux clusters at Supercomputing and
  Visualisation Unit (SUV), Singapore-MIT Alliance
  (SMA) and Faculty of Engineering connected to
  form a Grid testbed
• Linked to Linux clusters at NTU, IHPC and BII
• Globus Toolkit installed on all clusters
• Packages
• Condor, PBS, Sun Grid Engine

Courtesy of Tan Chee Chiang, NUS
9
ALICE (Adaptive and scaLable Internet-based
Computing Engine)
National University of Singapore

• Middleware
• lightweight ease of setup, use maintenance
• portability - platform independence
• generic infrastructure communication support
• maximizes throughput via job-parallelism
• maximizes performance via (Java)
  object-parallelism
• anonymity, security accountability
• performance (QoS) scalability
• Grid Programming
• ease of grid application development
• template-based programming hides complexities
  of parallel programming

Courtesy of A/P Teo Yong Meng, NUS
10
National University of Singapore
CRAY SV1 8 x CPUs GRID Host
NUS BioGRID
SGI 3200 Irix GRID Host
Linux P-III 800MHz GRID Host/GIIS/MDS
Sun UltraSPARC Solaris GRID Host
Linux cluster 14xP-III 800MHz GRID Host
(Sun GridEngine)
http//www.bic.nus.edu.sg/biogrid
Courtesy of A/P Tan Tin Wee, NUS
11
Asia Pacific BioGRID (as at 2002)
National University of Singapore
CrayMN
CrayJapan
Stanford
Kasertsart
UTM,UKM,USM

• NUS

UniMelb
Monash
ANU
http//www.apbionet.org/grid
Courtesy of A/P Tan Tin Wee, NUS
12
Institute of High Performance Computing

• Research Focus
• Middleware development (with NTU)
• Directory services, execution management services
• Projects
• Infrastructure implementation (KISTI (Korea))
• Grid Enabled Computational Electro Magnetics
  (British Aerospace, Cardiff University, Swansea
  University HP)
• Environment to support simulations (such as CFD,
  CME, CEE, MEMS)
• HPC Resources
• IBM p690 Regatta, 13 17 CPU PIII Beowulf
  clusters, etc

Courtesy of Dr. Kurichi Kumar, IHPC
13
IHPC GECEM
Institute of High Performance Computing

• Aim
• Use develop grid technology as an enabler of
  large-scale globally distributed scientific
  engineering research
• Focus
• Collaborative numerical simulation
  visualisation for Electromagnetic Application
  between UK Singapore
• Areas
• Grid Deployment establishing a Virtual
  Organisation
• Development of a Grid-enabled Problem Solving
  Environment
• Grid Services for Mesh Generation Manipulation
• Secure Remote Execution
• Grid based Collaborative Visualisation
• Evaluation Exploitation of Grid based Computing

Courtesy of Dr. Kurichi Kumar, IHPC
14
Institute of High Performance Computing
Fast Algorithm for Very Large Scale
Complex Aircraft Electromagnetic Simulation

• Project Focus
• Development of fast computational algorithms to
  accurately predict RCS for fighter
• Source codes are developed by IHPC
• Simulation was carried at IBM P690 with 32 CPUs

Simulation model for current distribution on an
aircraft
Courtesy of Dr. Kurichi Kumar, IHPC
Distribute structure of FMM
15
SER-Grid
Institute of High Performance Computing

• MPI application for MEMS division (Incompressible
  Navier-Stokes Solver)
• The Grid testbed for this experiment includes 2
  Linux clusters from IHPC, 1 Linux cluster from
  NUS 1 Linux cluster from NTU
• Globus Toolkit MPICH-G2 are used
• Resource management systems including Condor, PBS
  Sun Grid Engine are installed in these clusters

Micropump Simulation On the Grid
Courtesy of Dr. Kurichi Kumar, IHPC
16
Genome Institute of Singapore (GIS)Grid Projects
(in progress)

• SNPs
• The GIS is involved in building an integrated
  SNPs database taking information from 4 public
  SNPs databases and then aligns the information to
  a common genome sequence. This will help to
  identify SNPs which overlap more significantly
  provides a platform to target markers build
  primers to further identify the novel SNPs
  their relationship to gene function.
• Involves
• 3 million plus BLAST jobs
• Continuous updating reworking of the data as
  both the Human Genome map available SNPs change
• Requires highly parallel computing facilities

Courtesy of Heidi Dowst, GIS
17
Genome Institute of Singapore (GIS)Grid Projects
(in progress)

• Protein-Protein Interactions Database (PPDB)
• Building a system to help predict protein-protein
  interactions by integrating many tools data
  from several sources. We use information from
  domain fusion, text-mining, experimental
  databases, phylogenetic profiling to construct
  putative protein-protein interaction networks.
• Involves
• Combining numerous public databases from around
  the world
• Extensive pattern matching
• Use of Grid tools for the collection of data
  clusters for data mining

Courtesy of Heidi Dowst, GIS
18
Genome Institute of SingaporeClinical Research
Grid

• Seeks to link medical research organisations
  throughout Singapore by an operational system
  that can integrate tissue collection medical
  data while preserving patient privacy for the
  purpose of improving the national health with
  biological research
• Interacts with
• National registries that tracks the outcome of
  specific diseases important to the Singapore
  population
• Public Hospitals
• Research Institutes
• Tissues banks
• For the purpose of accessing, processing,
  archiving, distributing targeted tissues.

19
The National Grid Project
20
The NGP was formed in Sept 2002with the vision

• to facilitate the seamless use of an integrated
  cyber infrastructure in a secure, effective
  efficient manner to advance scientific,
  engineering bio-medical RD,
• with the longer term goal of transforming the
  Singapore economy using grid

21
Organisation Structure
MTI (ASTAR, JTC RICs)
MINDEF (DSTA, DSO, defence contractors)
MITA (IDA)
MOH (Hospitals)
MOE (Schools)
NUS, NTU and Polys
National Grid Project Steering Committee (NGP SC)
National Grid Project Office (NGPO)
Facilitates and co-ordinate activities
National Grid Operations Centre National Grid
Competency Centre (not formed yet)
Life Sciences Grid Virtual Community
Physical Sciences Grid Virtual Community
22
NGP Office(we are only 23-day old!)

• Set up on 1 Jan 2003
• Functions
• Secretariat to NGP Steering Committee
• On behalf of the Steering NGP Committee
• Facilitates co-ordinates activities of Working
  Groups Virtual Grid Communities
• Co-ordinates Grid Activities in Singapore
• Develops NGP Master Plan
• Acts a catalyst to achieve the NGP vision

23
The NGP Masterplan/budget is expected to be
drafted in April 2003

• Cyber-Infrastructure for R D
• Preparing Singapore for an Grid-enabled economy

24
Acknowledgements

• A/P Francis Lee (ebslee_at_ntu.edu.sg), School of
  Computer Engineering, NTU
• A/P Teo Yong Meng (teoym_at_comp.nus.edu.sg),
  Computer Science Department, NUS
• A/P Tan Tin Wee (tinwee_at_bic.nus.edu.sg),
  Bioinformatics Centre, Dept. of Biochemistry, NUS
• Tan Chee Chiang (ccetancc_at_nus.edu.sg), Computer
  Centre, NUS
• Dr. Kurichi Kumar (kkumar_at_ihpc.a-star.edu.sg),
  Institute of High Performance Computing
• Heidi Dowst (gishd_at_nus.edu.sg), Genome Institute
  of Singapore