The National Grid Project Office is aiming to create a permanent inter-grid between National University of Singapore (NUS), Nanyang Technological University (NTU), Institute of High Performance Computing (IHPC) and the Singapore Massachusetts Institute of Technology Alliance.
The Pilot Platform Grid project aims to create a more permanent grid. Targeted to be ready at the third quarter of this year, the Pilot Platform Grid will be running live applications on a more permanent basis.
“We hope to start small and build success stories,” said Lee Hing Yan, deputy director, National Grid Project Office. “When people find value in it, we will expand on it.”
Lee was speaking at the first Physical Science and Grid Computing Symposium which was held recently. The National Grid Project Office was set up on Jan. 1 2003 to facilitate the long-term goal of transforming the Singapore economy using grid computing.
Presently, there is an inter-grid between NTU, IHPC and NUS known as the Science and Engineering Research grid (SER-grid). This is a test bed grid used for experimental purposes such as trying out different middleware. “However, this inter-grid is not always on,” admitted Lee.
According to Professor Teo Yong Meng, the understanding of grid systems is already relatively well developed and has leveraged on knowledge gathered in the past for distributed computing. However, he said, the grid programming model still has some limitations. For example, it is not clear how debugging of grid software should be done.
Participants at the symposium agreed that there is a need to design better middleware and tools for engineers to use, and to create a more powerful and user friendly interface. More research and development is also needed in the areas of collaborative visualization and to develop new techniques to map parallel applications in grids.
For grid computing to have a prominent role in the commercial world, it must also first prove that it brings economic benefits, said IHPC professor Kurichi Kumar.
“Sharing is a dirty word,” said Teo. It gives the connotation of being free which does not quite gel with the commercial world. Grid computing, according to Teo, is meant to take advantage of under-utilized resources and not “free” as in without costs.
Benjamin Khoo, IT Availability Specialist at IBM and co-author of an IBM Redbook on high performance computing, said one way to commercial viability is to have the grids exposed as services for which users will pay or earn, depending on whether they are providing the services or consuming the services.