Nguyễn Quang Hùng

Enabling technologies in high-speed communication, CPU-production, Unix/Linux operating systems today have made clustered systems become one of mainstreams of parallel and distributed platforms for high-performance, high throughput and high-availability computing. The Unix/Linux clustered systems, were connected by thousand of similar workstations, are popular used to run large applications. Such a kind of large applications is the kind of Vietnamese semantic web applications, in which hundred thousands of entities in a knowledgebase are accessed and many coordinated complex processes will be done. In this paper, a high performance computing (HPC) clustered system for semantic web application, named SWG, are proposed to run Vietnamese semantic web applications. The SWG system is used to (1) store and query Vietnamese knowledgebase by Sesame API, (2) annotate Vietnamese web pages, and (3) access indexed documents by Lucence.

Grid Computing promises a revolution both science and social. We can be changed thinking about software and hardware by the Grid computing. However, developing a computing system and gridenabled application based on case by case. Although, there are some proposed standards such as Grid Architecture, OGSA for integrating distributed systems by grid services. This paper proposes a building common virtual Application Programming Interface (API) layer that can apply for e-science or ebusiness applications. Each virtual operation can be implemented by group of grid services that run on grid nodes. To prove that the virtual API layer can be implemented, this paper also describes about a grid environment called BK-GRID that is built on Globus Toolkit 3. The BK-GRID has eight grid-nodes. Each grid node is a PC Pentium IV that has RAM memory is 256 Mbytes.

Grid computing is a new computing infrastructure and Globus Toolkit (GT) is a de-factor grid middleware to build grid. However, the Globus Toolkit’s Monitoring and Discovery System (MDS) do not support a resource discovery service well. So this paper presents a user-driven grid resource discovery service. Users can describe resource requirement specifications by extended Job Specification Description Language (JSDL), in which each resource is given a range value with a lower bound and an upper bound. Our discovery service is Web Service Resource Framework (WSRF), Java multi-thread grid service. In addition, the paper introduces the resource set-matching and ranking algorithm.

Grid Portal is a easy way for user can submit job into a Grid infrastructure and monitor submitted job status. However, there are not a generic portal that can taylor for all Grid infrastructure, for example the EDAGrid@HCMUT (www.edagrid.hcmut.edu.vn). Therefore, we have proposed a new portal for the EDAGrid@HCMUT.