Problems Arise in $1 Million Computer System

Two years ago, a group of UC Irvine professors led by Charles Zender, assistant professor of earth system science, put together a proposal for the development of a computer modeling system that would allow students and faculty to study the atmosphere, ocean and land systems together, which had previously been possible only on a small scale.
As a result, the Department of Earth System Science received a ‘coupled climate model,’ a $1 million computer system which allows the professors to study the global climate. However, one year after the new system was installed, some of the components of the system are still not working as they were intended.
The system, called the Earth System Modeling Facility, consists of eight p655 machines with eight 1.5GHz Power4+ processors and one p690 with 32 1.7GHz Power4+ processors.
The hardware was supplied by IBM and the computer program was created by the National Center for Atmospheric Research in Boulder, Colo. Some modifications to the system were made on the UCI campus.
With this program, students and professors are able to research the effects of humans on the earth. By undertaking this development, the scientists hope that the ‘small-end supercomputer’ will help put UCI on the map as a preeminent center for climate research.
The program does work, for the most part. Several students have already done research and published papers using the system and many more are currently using it for research.
For the first three months, the system was operable, but was not used because the department did not fully understand the technology with which it was working.
Zender then went to the National Center for Atmospheric Research in Boulder, Colo. to learn how to use the program correctly. Since then, the system has worked almost exactly as planned.
Currently, however, the main problem with the system is lack a of storage space for the information generated from the program.
When the Department of Earth System Science purchased the hardware from IBM, it bought storage disks that were large enough to hold all the information it needed. It was later discovered that only about 10 percent of the disks work. This is enough to store small projects, but nowhere near enough to hold the information for its biggest planned project, which is the model of the world 300 years in the future.
‘I am disappointed that the disk doesn’t work because every [part of the system] has to work together to answer the big questions that we want to ,’ Zender said.
The Department of Earth System Science is presently trying to contact IBM in order to either receive new disks or to be reimbursed for the money it paid for them.
Though it has not yet heard from IBM, the department is hopeful that it will receive a response within the next few weeks.
Other than these storage problems, the system has been working as expected and has provided professors with a powerful research tool. Keith Moore, assistant professor of earth system science, has been doing research using the coupled climate model and said that if he had been using a regular computer, the information he gets in one day would take months to process.
‘It has been working very well,’ Moore said. ‘My only complaint is the storage problems and how that affects long term research. IBM needs to fix it one way or another. Other than that, the program is very fast.’