Recent Advances in China Grid Project Hai Jin

Recent Advances in China. Grid Project Hai Jin Huazhong University of Science and Technology Wuhan, China hjin@hust. edu. cn

Nation-wide Grid Projects in China • China National Grid – CNGrid (Ministry of Science and Technology ) • China Education and Research Grid China. Grid (Ministry of Education) – • China Science Grid Project (National Science Foundation) v. Two group efforts working on the grid middleware platform Ø Group 1: ICT/CAS, HUST, Peking U. , Tsinghua U. , Xi’an Jiaotong U. , and Shanghai Supercomputer Center (CSGrid project) Ø Group 2: Beihang U. , Peking U. , Tsinghua U. , NIC/ CAS, NUDT (China Research and Development Environment Over Wide-area Network, CROWN) 2 2

China. Grid in a Nutshell • China Education and Research Grid • Funded by Ministry of Education • As the pilot grid application supported by National 863 Hi-Tech R&D program • Based on CERNET (China Education and Research Network) • First Phase v From 2003 -2005 v 12 key universities as initiative v More than 6 Tflops w/60 TB v 20 key universities now 3 3

About CERNET • Second largest IP backboned network in China v 36 Po. Ps cover all the province capitals v The network backbone is 2. 5 -10 Gbps DWDM • Cover more than 1, 500 universities/colleges in about 200 cities v About 20 million users v Many universities have their own campus network in China 4 4

State-of-the-art of CERNET-1 5 5

China. Grid (The First Phase) 6 6

China. Grid (till now) 7 7

China. Grid Mission • Campus grid platform • Common platform for China. Grid • Grid application platform and representative grid applications v. Image processing grid v. Bioinformatics grid v. Course on-line grid v. Computational fluid dynamic grid v. Large scale information processing grid 8 8

• Campus Grids and their Computing Power and Storage Capacity HUST： 0. 8 Tflops, 5 TB • ZSU： 1. 3 Tflops, 10 TB • THU： 1. 3 Tflops, 20 TB • PKU： 1. 5 Tflops, 10 TB • BUAA： 0. 7 Tflops, 10 TB • SCUT： 0. 9 Tflops, 20 TB • SJTU： 0. 5 Tflops, 9 TB • SEU： 1. 0 Tflops, 19 TB • XJTU： 0. 1 Tflops, 2. 5 TB • NUDT： 0. 2 Tflops, 5 TB • NEU： 0. 8 Tflops, 5 TB • SDU： 1. 3 Tflops, 18 TB • NWPU： 0. 4 Tflops, 2 TB • ZJU： 0. 6 Tflops, 3 TB • FUDAN： 1. 8 Tflops, 6 TB • TONGJI： 0. 2 Tflops, 1 TB • USTC： 1. 2 Tflops, 5 TB • UESTC： 0. 3 Tflops, 2 TB • RUC： 0. 1 Tflops, 0. 5 TB • LZU： 0. 3 Tflops, 1 TB China. Grid：>15 Tflops, >150 TB 9 9

Layered Infrastructure of China. Grid ZSU PKU SJTU XJTU NEU HUST SCUT BUAA NUDT Remote Image education processing grid THU Fluid dynamics grid Bioinformatics grid SEU SDU Massive information processing grid China. Grid Supporting Platform (CGSP) High performance computing environment (campus grid) 10 10

What’s CGSP? • China. Grid Supporting Platform, a grid middleware for China. Grid • Follow the WS-Resource Framework (WSRF) and Open Grid Service Architecture (OGSA) • Beta version was announced last October at GCC’ 04 in Wuhan • CGSP 1. 0 was released Jan. 10, 2005 • Developed by the joint working group from HUST, THU, PKU, BUAA and SJTU 11 11

Status of CGSP 1. 0 • CGSP download more than 300 times • Support three major grid applications of China. Grid • Installed in 12 universities as campus grid environment 12 12

China. Grid Supporting Platform Portal: Grid entry for submitting & monitoring job, querying resources’ info, user management and accounting etc. Grid Developing Environment: a set of toolkits including portal development tools, resource encapsulation tools, programming tools and job generation tools etc. Information Center: the manager of resource & service information Uniform Management: a set of managers including job manager, data center, domain manager and service container Security: Identity authentication and mapping, service and resource authorization, secure transferring etc. 13 13

GT 3. 9 and CGSP Proxy (CGSP) GSI WS-Security Data Center (CGSP) New Grid. FTP Meta. Data-Man (CGSP) RFT Job Manager Pre-WS (CGSP) GRAM Work. Flow (CGSP) WS-GRAM CSF (contribution) CAS RLS Simple. CA OGSI-DAI Portal SDK (CGSP) Authz Framework XIO Programming Toolkits (CGSP) Security Data Management Resource-Index (CGSP) MDS 2 Hyper. Service WS-Index (CGSP) Inter. Domain (CGSP) Topologic (CGSP) py. Globus (contributed) Information Services JAVA WS Core (WSRF) C WS Core (WSRF) Python Bindings (contributed) Remote. Deploy (CGSP) Hot. Deploy (CGSP) WS Core 14 14

CGSP in Detail Portal development tools Resource encapsulation tools Programming tools (Grid. RPC/Grid. PPI) Job generation tools Management tools Installation tools Grid Developing Environment Service management Resource management Fault-tolerance mechanism Domain info management User management Identity mapping Service matching Topology management Quality of service management Metadata description Status monitor Negotiation policy Information Center Job submission Job monitoring Work flow management Domain Manager Job remote-deploy Job scheduling Service support Job Manager Uniform access entry Metadata manager Remote&hot deploy Service monitor Node resource monitor Batch job service Base fault Service group GT 3. 9. 1 core Storage resource manager Single sign-on Notification Resource property Service Container Identity certification Proxy management Replica catalog Life cycle Certification authority Storage agent Data Manager Security management of service and container Resource access control Security 15 15

Working Flow of CGSP 16 16

17 17

AG 4 China. Grid：Grid. Asia 2005 18 18

http: //www. chinagrid. edu. cn 19 19

Bioinformatics Grid 20 20

Bio. Grid Applications • Protein target selection for rice genome • Multi-sequence alignment for ganoderma family • Gene joint for white mice • Cardiovascular disease research 21 21

Image Processing Grid 22 22

Image. Grid Applications • Three-dimension reconstruction of digital virtual human being (HUST) • Medical image diagnosis (ZSU) • Remote sensing image processing（NUDT） 23 23

CFD Grid 24 24

Pilot Applications based on CFD Grid • Shanhai Jiao. Tong University: Aircraft genetic optimization design system • Xi’an Jiao. Tong University: Parallel algorithm for molecular dynamics • Southern China University of Technology: Simulation and optimization techniques for metal powder figuration process • Northeast University: Large scale geological disaster forecasting system • Southeast University: Computational electromagnetism applications in airplane and automobile design 25 25

Course On-line Grid 26 26

Course Resource Contribution • Number of courses: 233；Number of hours: 2566 27 27

Usage Survey User IP distribution（2004/08/27 -2005/01/16） Server Independent Total number of IP number HTTP GET PK 1 15359 5868376 SD 1110 128816 PK 2 264 10970 UST 9513 1093657 BY 2922 640255 XJ 3981 697464 BH 121 10383 SJ 528 265055 TS 818 125083 XB 1318 160240 DB 1358 At least 848936 so 696 many people 71957 LD 5101 618136 HZ 598 64299 DN ZS 9370 use course 1260541 online grid HN 387 Total number 705687 of access Numbers of independent IPs： 25, 807；Total number of HTTP GET： 12, 569, 855 28 28

Massive Information Processing Grid 29 29

University Digital Museum Grid (UDMGrid) 30 30

International Collaboration (1) 31 31

International Collaboration (2) • GPE from D-Grid with CGSP • Access Grid with USA and Singapore • Trusted Computing for Grid Security with HP • Global Grid Monitoring with HP • Also collaboration with IBM, Sun, etc • More collaboration with EU 32 32

Thanks! 33 33