Archived content
NOTE: this is an archived page and the content is likely to be out of date.
Grid Computing
2004-12 (Vol.40, No.2)
This special issue introduces Grid computing technologies emerging to enable more effective use of geographically dispersed heterogeneous computers thanks to various innovations of information technology (IT). Although there are many Grid issues to be solved, Fujitsu will continue its efforts and create a new Grid-based computing paradigm.2004-12 (Vol.40, No.2) Contents
- 1. Preface (27 KB)
- In 1965, Gordon Moore, one of the founders of Intel, predicted that transistor counts in chips would double every 18 months to 2 years; this is known as Moore's Law. Later, in 1998, George Gilder, the visionary author of "Telecom," stated that bandwidth grows at least three times faster than computer power. Therefore, according to Gilder, if computer power doubles every 18 months (as per Moore's Law), then communications power doubles at least every 6 months. Because of various innovations in information technology (IT), these predictions have come true, and as a natural result of the improvements in powerful computers and networks, Grid computing technologies have emerged to enable more effective use of geographically dispersed heterogeneous computers. ---[Kimio Miyazawa, Senior Vice President Fujitsu Laboratories Ltd.]
- 2. Fujitsu's Activities in Grid Computing (136 KB)
- Fujitsu has been actively involved in the research and development of Grid computing: a technology that enables the integrated and collaborative sharing of resources across administrative boundaries. Research on Grid computing was initially carried out in scientific computation projects, and Fujitsu has been participating in a number of such projects in Japan and Europe. As the technology became more mature, Fujitsu's R&D focus expanded to include business applications of Grid computing, with interoperability as a key issue. Therefore, Fujitsu has been increasingly involved in the standardization of Grid computing, especially technologies relating to commercial Grid systems. In addition to working on standardizing essential technical specifications, Fujitsu also holds a number of key posts in the Global Grid Forum, which is the premier standard-setting organization for Grid technology. Fujitsu has also been promoting the establishment of important standards in other bodies, for example, OASIS. In this paper, we describe key Grid computing projects that Fujitsu has been engaged in and some current technology-development activities. We also outline the technical fields that are expected to become significant in Grid computing in the future. ---[Andreas Savva, Takumi Ouchi, Hiro Kishimoto]
Science Grid
- 3. Overview of Japanese National Research Grid Initiative (NAREGI) Project (144 KB)
- The Japanese National Research Grid Initiative (NAREGI) is a 5-year project covering the period 2003 to 2007. The goals of the project are to 1) conduct R&D on high-performance and scalable Grid middleware to provide a future computational infrastructure for scientific and engineering research in Japan, 2) Grid-enable a specific target application, in this case nanoscience, to prove the usefulness of Grids in future scientific and industrial applications, and 3) perform experimental deployment of 100 TFLOPS-scale Grids based on the middleware and Grid-enabled applications that have been constructed. The National Institute of Informatics (in charge of Grid middleware R&D), the Institute for Molecular Science (in charge of Grid-enabling nanoscience applications), major universities, and industries will participate in this project. The Super SINET, an all-optical 10 Gbps and beyond backbone network for academic research, will be used as the network infrastructure for this project. This paper outlines the NAREGI project. ---[Kenichi Miura]
- 4. VizGrid: Collaborative Visualization Grid Environment for Natural Interaction between Remote Researchers (276 KB)
- Recently, as the borders between research fields have become fuzzy, there have been increasing opportunities for researchers in different fields to collaborate with each other. VizGrid is a joint research project between universities, government institutions, Fujitsu, and other industry members that was partially founded by the Ministry of Education, Culture, Sports, Science and Technology. This project aims to develop a new collaboration space in which three-dimensional (3D) images can be manipulated in a highly realistic virtual environment. This paper describes Volume Communication, which is an infrastructure we developed for the VizGrid project that can generate, communicate, and display functions of 3D image data. We also describe the technical problems in transmitting high-resolution 3D images based on the evaluation of the prototype system we developed. The collaboration space proposed in this paper is used to communicate common objects that configure a virtual environment and images of humans as 3D image data. The receiving side visualizes this data so that important functions, for example, eye-to-eye contact, can be realized to simulate a real face-to-face environment. ---[Ryuichi Matsukura, Koji Koyamada, Yasuo Tan, Yukihiro Karube, Mitsuhiro Moriya]
- 5. Construction of Japanese Virtual Observatory (JVO) (97 KB)
- Large amounts of high-quality astronomical data are obtained from the Subaru Telescope and other observatories located throughout the world, and this data is stored in worldwide locations. It is now more important than ever to have an observational database that enables astronomers to search for and collect valuable astronomical data from this vast collection and perform research through statistical analysis. The National Astronomical Observatory of Japan (NAOJ) has therefore started construction of a system called the Japanese Virtual Observatory (JVO) that will connect these observational databases via high-speed networks so they can be used collectively as a single virtual data archive. Fujitsu has been collaborating with the NAOJ to develop JVO prototype systems. To build the JVO, a virtual integrated environment for the distributed data will be required. To achieve this, three issues must be considered: 1) construction of a virtual data storage system, 2) efficient use of the vast amount of distributed data, and 3) the fragileness of the distributed environment. This paper describes and evaluates the JVO prototype constructed using Data Grid technology to ascertain the feasibility of a virtual integrated environment for the distributed data. ---[Yasuhide Ishihara, Yoshihiko Mizumoto, Masatoshi Ohishi, Kenji Kawarai]
- 6. CAD-Grid System for Accelerating Product Development (89 KB)
- Recently, there have been demands for greater functional diversity and higher performance in the information technology (IT) infrastructure and mobile communication products. As a result, the complexity and scale of product design in the product development phase have increased. In addition, development periods need to be reduced so products can be marketed sooner. These requirements have rapidly increased the need to efficiently simulate and analyze in order to achieve optimum designs and verification. Also, Grid computing technology is coming into practical use as a new way of using geographically distributed computer resources connected to a network. This technology is based on advances achieved in high-performance computing and broadband networking. The authors have constructed the "CAD-Grid" Grid computing system to quickly perform computationally intense simulations in the product development phase. This system has been used to simulate a mobile communication system and confirm its effects. This paper describes the Grid computing environment of CAD-Grid, its structure, the system simulation to which CAD-Grid was applied, and the application results. This paper also describes the future development plan for the CAD-Grid system. ---[Tomonori Yamashita, Takeo Nakamura, Hiroshi Noguchi]
- 7. Development of PIV Web Laboratory and Its Application (366 KB)
- Particle image velocimetry (PIV) is an experimental method of researching fluids that is employed in fields such as automobiles, aviation, civil engineering, and medicine. It is used as a complement to computer simulation. Three-dimensional analysis of PIV data requires a high-performance computer, large-capacity storage, and a powerful visualization engine. Also, researchers require an environment for collaborative PIV research using powerful computer resources. The PIV Web Laboratory (PIV-WL) was developed to meet this requirement. This system is based on UNICORE Grid middleware and has multi-job management and 3D remote visualization functions. We used this system to analyze the flow fields on delta wings and created a collaborative research environment in which many users can manipulate visualization results simultaneously by remote control. ---[Taiyo Maeda, Naoki Onishi, Yoshimasa Kadooka, Yoshio Tago]
- 8. RIKEN Super Combined Cluster (RSCC) System (147 KB)
- Recently, Linux cluster systems have been replacing conventional vector processors in the area of high-performance computing. A typical Linux cluster system consists of high-performance commodity CPUs such as the Intel Xeon processor and commodity networks such as Gigabit Ethernet and Myrinet. Therefore, Linux cluster systems can directly benefit from the drastic performance improvement of these commodity components. It seems easy to build a large-scale Linux cluster by connecting these components, and the theoretical peak performance may be high. However, gaining stable operation and the expected performance from a large-scale cluster needs dedicated technologies from the hardware level to the software level. In February 2004, Fujitsu shipped one of the world's largest clusters to a customer. This system is currently the fastest Linux cluster in Japan and the seventh fastest supercomputer in the world. In this paper, we describe the technologies for realizing such a high-performance cluster system and then describe the implementation of an example large-scale cluster and its performance. ---[Kouichi Kumon, Toshiyuki Kimura, Kohichiro Hotta, Takayuki Hoshiya]
Business Grid
- 9. Business Grid Computing Project Activities (169 KB)
- The recent global diffusion of the Internet has increased the importance of e-business and created new opportunities and challenges. One effect is that the business cycle, from planning to development to operation, is becoming shorter and shorter. Therefore, there is increased pressure on companies and public institutions to construct and manage reliable and flexible application systems at low cost. The Business Grid Computing project is aiming to address these new requirements by developing Grid technology based on virtualization, autonomous control, and unified management of IT resources. The project is based on the Open Grid Services Architecture (OGSA), which is a next-generation standard architecture unifying Grid and Web services technology. To improve the interoperability of Grid systems, the project is also promoting international standardization of the technologies it develops. This paper describes the technologies developed by the Business Grid Computing project and also gives an overview of the standardization activities. ---[Andreas Savva, Toshiyuki Suzuki, Hiro Kishimoto]
Grid Middleware
- 10. Grid Middleware for Effectively Utilizing Computing Resources: CyberGRIP (92 KB)
- Various research and development activities regarding Grid computing technology have recently been promoted. This technology integrates heterogeneous computing resources that are geographically dispersed and virtualizes them as a single computer system. However, there have been very few reports in the business world about Grid computing becoming successful or practical because most Grid computing research is carried out in government science projects. Fujitsu has developed CyberGRIP, which is a Grid middleware system for applying Grid computing to practical corporate simulations and verifying the effectiveness of Grid computing. This paper clarifies the needs and problems of these corporate simulations and describes how CyberGRIP is solving the problems. This paper also describes in-house verification using the CAD-Grid system. ---[Akira Asato, Yoshimasa Kadooka]
- 11. FSE Grid Middleware: Collaborative Grid Environment for Distributed Computing (229 KB)
- While traditional distributed computing technologies provide critical building blocks and frameworks for Grid application development, distributed computing is not concerned with and does not address the following: large-scale and dynamic resource sharing, frequently stringent performance requirements, large resource needs, and the multidisciplinary nature of Grid applications. The goal of the FSE Grid middleware project is to develop the knowledge and technology base required to support application execution in a collaborative environment for distributed computing along with application development strategies to make FSE Grid middleware accessible to ordinary scientists, engineers, and software developers for problem solving. ---[Roger Henri, Pierre Lagier, Didier Plaindoux]
- 12. Explicit Trust Delegation: Security for Dynamic Grids (77 KB)
- This paper addresses the issue of how to build dynamic Grids without using the non-standard proxy extensions that cause concern within the security community. The approach allows commonly available security tools and libraries to invoke requests (on the client side) and respond (on the server side) using only well-established security protocols. This description is made in the context of the UNICORE Grid infrastructure. UNICORE is known to have a strong, respected security model, but at the cost of not supporting some dynamic Grid capabilities. The discussion shows how UNICORE could be enhanced using Explicit Trust Delegation to provide dynamic capabilities, hitherto only possible in Grids supporting a proxy-based security model. We show how this approach provides a smooth migration path to proposed work on Virtual Organizations. ---[David F. Snelling, Sven van den Berghe, Vivian Qian Li]
