e-Science Development of Taiwan Eric Yen & Simon Lin ISGC, March 2011
Outline • Extending Regional Production DCI • Conducting e-Science Collaborations • Life Science, Earth Science, Environmental Changes, Social Science and HEP • Technology Development • Building New generation DCI and Operation Technology • Application Technology • Service-Oriented Computing & Cloud • Dissemination, Training and International Collaboration 2
e-Science Infrastructure in Asia *+,-.-/.)&0)!"#$%&'(),12)34&5.)61)7"89!) #)!!!!!" #$!!!$!!!" • 38 sites in 15 countries ($'!!$!!!" • > 1,800 Users #!!!!!!" ($&!!$!!!" • Average availability > 90% after Nov. 2010 ($%!!$!!!" • 15,000 Cores, 5 PB Disk, 4 PB Tape ()!!!!!" ($#!!$!!!" • 62K Jobs/day, 80K CPU-Hrs/day !"#$%&'() ($!!!$!!!" • EUAsia, LHC experiments, Biomed, etc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
e-Science Networking in Asia Pacific Region ! 6+ PB data in/out ASGC in 2010 ! 11 Gbps reached PH-ASTI- VN-IFI- VN-IOIT- TH- MY-UM- MY-UPM- MY-MIMOS PK-PAKGRID PK-NCP LIKNAYAN PPS HN NECTEC CRYSTAL BIRUNI-01 TH-HAII APAN-JP JP-Tokyo- VN-IOIT- LCG2 IP Transit KEYLAB SINET JP-KEK1 ITB-ID NL JP 2.5G 10G CERNET WIDE iHEP-CAS JP-KEK2 2.5G HK TW US JP- CN-SDU- 5G HIROSHIMA- LCG2 WLCG2 I2 / GN2 622M CSTNET HK-HKU 10G IN-TIFR SG TWAREN/ CN-BEIJING- KREONET IN- LCG2 TANET VECC1 KR-KISTI- AARNET IN- GCRT VECC2 TW- TW- TW-FTT AU-ATLAS ASGC TW-NIU KR-KNU NTU NCUHEP NTCU ALICE EUAsiaGrid CMS 4 ATLAS Sites Sites Sites Sites
Resource Status • All resources are integrated and managed by Grid system. • Operated and managed by ASGC. CPU Disk Tape Resource Groups (TB) Inter-Conn User Groups Cores (TB) 4,504 4,660 4,020 WLCG, TWGrid Ethernet E-Science, HPC, Grid Earth, Env. Changes, 0 10G Ethe + IB and Cloud EUAsiaGrid, 5,640 700 (DDR/ QDR) Applications Astronomy & HPC Cloud, Other e- 4,416 470 0 Ethernet Science 5
Monitoring Tools/Alarm System at ASGC Ganglia ! Weathermap ! MRTG ! 6
System Optimization • Performance, Cost, Energy Saving, Early Warning and Automation • Storage System and Data Management • Deploying higher density disk array with large bandwidth • 24x2TB array " 96x2TB array • #storage servers reduced, and 10Gb Ethernet and 8Gb fibre interface equipped • Castor and DPM performance tuning: from array controller to DPM/Castor and intermediary services are explored. • Merging ATLAS storage class to DPM • Reduce data transmission between ASGC and TW-FTT • Castor takes care only Tape-required data services • Distributed file system • Computing System • Networking: from DC to international connection • DBMS Architecture and Services 7
Stress Test and Details at “Operation & Management” Session, Performance Tuning 1600, Mar 23 8
Smart Center • Power Efficiency • Increase power efficiency by eliminating the use of UPS • UPS reduces power efficiency by 30 percent. Among them, 10 percent is in the form of heat that has to be carried away. • Thermal Efficiency • Apply space technology to heat conduction of the data center to increase thermal capacity • Intelligent Monitoring & Control • Analyzing long term data allows us to build models that can assist us in operating the center intelligently " early warning and automation 9
Cloud Technology • VM Management Framework • Data and Storage Virtualization • Application Platform Management • Easy software provisioning of identified applications • WLCG, targeted e-Science applications, MapReduce + Hadoop, ! • Brokering Services • resource level and service level • Monitoring and Accounting • Interoperability (between cloud/grid and among component/layers) • Standardization 10
11
Exercises and Use Cases • VM deployment • Benchmarking on cases from 1s, 10s, 100s, to 1000s VMs at a time • Minimize the VM image size, either for general purpose or customization • VM live migration • Through Global File System • gLite Work Node on-demand • VO-based resource policy implementation • Evaluate the impact to resource utilization and best practices • Employ P2P solutions for data transmission/location • Monitoring & Accounting – • leveraging current WWG services and add on those missing components • Nagios-based framework 12
:#7.6,G(62H):G::H):GI)A,.)J,/6=6+,-1>) 8<>6&1,=)!&==,5&(,-&1),12)K(62>61>) Enabling Grids for E-sciencE 7.6,)A6+D)+D<)L&(=2) :;/<==<1+)"(&>(<..) :;/<<2<2) :;?</+,-&1.)@) 7*)(&=<)A,.)B<C)0&()+D<)?(&E</+).'//<..F) Computational Chemistry Social Science Bioinformatics and Biomedical High Energy Physics ( W Mitigation of natural disasters � www.egi.eu EGI-InSPIRE RI-261323
e-Science Collaborations in Asia Discipline Applications Partners Going DG HEP ATLAS, CMS, ALICE, BELLE, CDF, GEANT4 TH, TW, CESNET, INFN X BioMedical Virtual Screening for Drug Discovery – Avian MY, TW, VN, CESNET, X Flu, Dengue Fever INFN Pandemic disease analysis VN, FR Bioinformatics Grid enabling phylogenetic inference SVM Parameter optimization for prediction of Caspases SG, TW, VN, CESNET, Genome search to identify T3SS effect X INFN Autodock ligand-receptor docking X Complex diseases studies Earth Science Disaster Mitigation on Earthquake ID, MY, PH, TH, VN, TW, X CESNET, INFN Comp Chemistry Chemical compound property analysis TH, TW, CESNET X Climate Change Weather simulation, sea level rising ID, PH, TH, VN, TW Social Sci. Social Simulation TW, UK X 14
Application Repository Application Status: S1 (in consideration), S2 (running but not ported to gLite yet), S3 (ported to gLite, unavailable in EUAsia VO), S4 (available in EUAsia VO), S5 (ready for production)
EUAsiaGrid Portal • Convenient access to grid infrastructures for individual users • Provides, through the portal interface, support to: • Submission of jobs • Specific forms for individual applications • Helping to prepare the job description and input data • Data management • Allow sharing with other users • Job Monitoring • Life Sciences – Autodock 4 , Beast, Blast, Gromacs, MrBayes, Muscle, Prodist – GVSS * • Earth Science: Earthquake * • Weather Simulation: WRF * • Statistics: R • Other User Defined Applications
Exemplar Applications 17
Grid Virtual Screening Service by AutoDock • One-click job submission ! Submit the docking job to the Grid with just one click ! • Visualize your job status ! SG + DG • View the best conformation of a • Generate the histogram with a given energy simulation ! threshold ! 18
GVSS � " 2006 ! • GAP release ! • Avian flu (DC2) / DCR drug screening (2.4M docking, 137+ CPU-years) " CNU/KR for wet lab test ! 2007 ! • Dengue fever (NS3) / CDI drug screening (300K docking, 4167 CPU-days) ! 2008 ! 2009 ! • GVSS release ! • Dengue fever / ZINC, ChemBridge drug screening ! • Antibiotics / GRC drug screening ! 2010 ! • Compound Profiling ! 19
e-Science for Earthquake Collaborators: PH, VN, TW, ID, MY, TH Disaster Mitigation Seismic Sensor Networks Local Sensor & High Observation Data Resolution Source & Global/Regional Sensor Rupture Data Process Analysis Fast Reporting System Archive Ref. Historical Events Data Forward Simulation & Event Archive Construction on Risk Analysis & Earthquake Data Reduction Grid Center (SeisGrid)
Seismogram Simulation Services 1. Location and Tomography 2. Epicenter Data Preparation Model Selction 3. Choose Position for Seismogram 4. Seismogram Access & Visualization
Recommend
More recommend
