1. Scientific Data Grid and e-Science
Kai NAN
Computer Network Information Center (CNIC) Chinese Academy of Sciences (CAS)
14 Dec 2006

2. Outline Background Scientific Data Grid (SDG) Prospect
Scientific Database (SDB)
CSTNet & GLORIAD
Scientific Data Grid (SDG)
Middleware
Visualization
Prospect
CAS e-Science

3. Background Scientific Data Grid (SDG)
Built upon the mass scientific data resources of the Scientific Database (SDB) of CAS
Led by CNIC; dozens of institutes of CAS participated
The vision of SDG is to take valuable data resources into full play by benefiting from advanced information technologies, in particular, the Grid technology.

4. Scientific Database (SDB)
SDB is a long-term project since 1983, in which there are multi-disciplinary scientific data accumulated through the course of science activities in CAS.
many institutes involved
long-term, large-scale collaboration
data from research, for research

5. SDB status 45 institutes across 16 cities 503 databases
16.6TB total volume

7. CSTNET Bandwidths Backbone 10G MAN link 1G WAN link 2.5G/155M
Cover more than 20 provinces, 100 institutes, 1,000,000 end users
12 Subcenter in China
Now CSTNET has twelve sub center distributed in China, like shanghai, wuhan, shenyang, xian, and other palce This picture shows us all of the nodes that CSTNET can reach in China.

8. CSTNET Internet Connections
622M
2.5G
155M

9. CSTNET in GLORIAD GLORIAD CSTNET Global Ring 10G 11 partners
GLORIAD_CN
Build HKOEP

11. SDG Supported by Chinese Academy of Sciences (CAS)
10th Five-year Informatization Program ( )
11th Five-year Informatization Program ( )
Ministry of Science & Technology of China (MOST)
863 Program ( )
863 Program ( )
National Science Foundation of China (NSFC)
Network-based Science and Research Environment (aka. NSFC e-Science) ( )

12. SDG Milestones
In 2000, the Scientific Database (SDB) project renewed fund by CAS 10th Five-year Program
In March 2001, proposed “Scientific Data Grid”
In October 2002, SDG joined the China National Grid (fund from MOST)
In Nov 2003, SDG Middleware v1.0 released
In July 2004, SDG got fund from NSFC
In Sep 2004, SDG renewed fund from MOST
In Oct 2004, DeepComp 6800 for SDG installed
In Nov 2004, SDG Middleware v2.0 released
In Aug 2005, SDG Middleware v2.1 released
Now, we’re working for SDG in 11th Five-year Program

13. Scientific Database (SDB) & Scientific Data Grid (SDG)
45 institutes participated
503 databases
16.6 TB
236-CPU Superserver (1TF)
20TB Disk Array
50TB Tape Library
VizWall & Access Grid

14. Requirements and SDG
How to FIND the data I want from hundreds or thousands of databases
How to ACCESS large-scale, distributed and heterogeneous scientific data uniformly and conveniently
How to make sure all this goes always in a SECURE and proper way

15. SDG Software Architecture

16. Data Access Service (DAS)
Uniform Access Interface (read-only)
Rich metadata
Easy publishing on web
flexible configuration and extensibility

17. DAS modules Data Access Interface Virtual Database Physical Database
MappingBuilder
DataView

18. SDG Services Chinese Ancient Astronomical Phenomena Database
Emperor’s name: Kang Xi
Keyword: 日食(solar eclipse)
天象(astronomical phenomena)
Chinese Ancient Astronomical Phenomena Database
DataView Service

20. MappingBuilder & Dataview

21. SDG Today

22. sdb6800 Superserver 59 nodes /236 CPUs
official service started in Apr. 2005
node usage 79.7% storage usage 87% (by Sep 2005)

23. SDG Storage System

24. Visualization System

25. portal.sdg.ac.cn

26. SDG CA SDG security infrastructure Accredited by ApGridPMA & IGTF
The subordinate CA of CNIC Grid CA
Accredited by ApGridPMA & IGTF
SDG CA Repository
CP/CPS, Introduction, Manual
Type of certificates
Person Certificate: Xu,DC=SDG,DC=Grid,DC=CN
Host Certificate: CN=sdg6800.sdg.ac.cn, DC=SDG,DC=Grid,DC=CN
Service Certificate: CN=DAS/sdg6800.sdg.ac.cn, DC=SDG, DC=Grid, DC=CN

27. CA Software Suite SDG CA V3.1 Certificate Utility V2.0
Based on the OpenCA V
Localization and Improvement
Certificate Utility V2.0
i18n

28. Applications High Energy Physics Astronomy Biology Geosciences …
Cosmic Ray Data Processing (YBJ)
Astronomy
China Virtual Observatory
Biology
Avian-flu Comprehensive Information Platform
Geosciences …

29. Tiled Display Wall @ CNIC
A Tiled Display Wall (TDW) is a large-scale high-resolution display
CNIC’s TDW uses a five by four grid of LCDs and 21-nodes cluster running Rocks (with CentOS installed).
1+20 cluster; 1Gigabit network for each node.
A Tiled Display Wall is a large-scale high-resolution display. It is composed of multiple projectors or LCDs whose images are projected side-by-side to create a tiled mosaic of imagery.
For some situations, an application runs on a single machine and a software or hardware distribution system divides up the screen image and distributes it out to the individual projectors, so that the resulting image appears as a single cohesive display. In others, a copy of the application runs on each machine, generating the imagery for the corresponding tile.
CNIC’s Tiled Display Wall is setup at 2005, with the support of Chinese Academy of Sciences. We use a five by four grid of LCDs and 21-nodes cluster. The Rocks is running on the cluster. As you may know, Rocks clusters soft is developed by SDSC rocks group and now is widely used in world-wide cluster and it is a convenient tool and software to setup a cluster. For the network, dual one gigabit network adaptors are equipped for each node and one Cisco switch is installed for the tiled display wall.

30. Visualization Software Package
Rocks Viz Roll – SDSC
DMX - Distributed Multihead X
SAGE - EVL/UIC
For the visualization Software package, two visualization system is installed in CNIC tiled display wall. One is rocks viz roll which is developed by SDSC group. Mason Katz and other engineers are involved in the project. Another one is SAGE visualization software, which is developed by Electronic Visualization Lab, SAGE group in University of Illinois at Chicago. As I know, KISTI also has setup a tiled display wall with SAGE visualization environment.

31. PyMol
This is a snapshot of Pymol application to show the avian flu molecule.

32. xnview
This is a snapshot of vnview image show application.

33. JuxtaView
This is a snapshot of JuxtaView. It is a image display tool which can piece lots of images together and give a large-resolution image showing. We also can use the client tool to control the image showing, including zoom in, zoom out, move left, right, down and up. In the snapshot, the total image size is more than six hundred Megabytes with more than by 8000 pixels. and just a part of image is showed on the tiled display wall.
The most important one for JuxtaView is that JuxtaView uses MPI parallel environment to accomplish distributed and parallel rendering and showing.

34. Collaborations PRAGMA EUChinaGrid ApGrid PMA / IGTF …
EUChinaGrid
Interconnection and Interoperability of Grids between Europe & China
ApGrid PMA / IGTF …

35. Prospect – CAS e-Science
CAS Informatization Program ( )
emphasis on Infrastructure Upgrade

36. Resources Lenovo 6800 Superserver Storage VizWall
Scientific Data (SDB)
Science Digital Lib (CSDL)

37. CAS e-Science Initiative 2006-2010
e-Science would be applications-driven
focus on implementation of e-Science Virtual Labs, the way for scientists to use
infrastructure may need refactoring

38. e-Science Virtual Labs
special meanings in the e-Science context
the key position in our e-Science framework
the core component to make e-Science a reality

39. vLabs Requirements
Infrastructure may be (almost) ready, but e-Science is not yet.
so many existing resources in place, but just a few could be brought into full play even now, with an advanced infrastructure ready.
bottleneck may be the gap between products by computer experts and end users of domain scientists
much more effort than expected to bridge this gap
Virtual Lab is proposed to be
a basic unit of research activity in the e-Science environment
the right user interface between scientists and their e-Science environment

40. Conclusion Remarks
Serving researchers in CAS with cyberinfrastructure is CNIC’s mission.
We have been building and running grid environments to support research activities better.
SDG and vLabs will play important roles in the CAS Informatization Program in which e-Science is the goal.
We look forward to further development and collaboration.

41. Thank you!