February 12, 2009
Information technology needs present challenge, interviews show
Meeting the information technology needs of the UW’s research leaders is going to challenge the University to provide the right kinds of resources and sufficient resources, not just in traditional infrastructure but in human expertise.
That’s the conclusion from detailed interviews with 125 research leaders. The results will be presented in a Catalyst Spark Session from 1 to 2:30 p.m. Friday, Feb. 20 in 220 Odegaard.
“No other university has conducted such a balanced study of top researchers’ information technology needs,” says Ed Lazowska, Bill & Melinda Gates Chair of Computer Science & Engineering, and one of the study’s leaders. “What we have found is a rich texture of IT needs, because these researchers are using computing in increasingly sophisticated ways.”
It used to be that complex simulations were the leading edge computational science research projects on most campuses. Now they have been joined by projects using arrays of inexpensive sensors and equipment such as desktop gene-sequencing machines that can generate a terabyte of data a day (a terabyte is 1,000 gigabytes or one million megabytes). Other data-intensive projects have recently come on line, such as the Large Synoptic Survey Telescope and the Large Hadron Collider, which will generate data at 30 times the rate of desktop gene sequencers. Locally, the Ocean Observatories Initiative, in which the UW is playing a central role, is similarly data-intensive.
Researchers are predicting that the next generation of data-centered computational science, commonly known as eScience, will be pervasive. To find out more about the implications of this change for the University, the Office of Learning and Scholarly Technologies joined the UW eScience Institute to conduct a study involving interviews with many of the UW’s top researchers across a range of fields and at various stages in their careers.
The issue on which there was the broadest agreement was the increasingly competitive funding environment. It follows that those universities which are more successful in providing the technology resources necessary for conducting research will be most successful in obtaining funding and keeping top researchers. Researchers also told interviewers that their work is increasingly interdisciplinary and multi-institutional, driven by the complex research questions being asked and by granting agencies’ preferences.
Because researchers are collecting increasing volumes of data, their needs for data management are growing. They need access to more storage, better backup systems and greater data security. In addition, the exponential growth in data poses challenges for researchers in the analysis phase. The number of people who have high level computational skills, as well as in-depth knowledge of that particular field of research, is generally very limited.
Many researchers said for the future they would need to call on more expertise in configuring and managing databases and computer servers. “Currently, we operate on an ‘every ship on its own bottom’ model, with very limited sharing of expertise across campus, which is also true at other universities,” Lazowska says. “This is not very efficient and requires many researchers to reinvent the wheel.”
Research projects are not simply generating more data, they are generating a greater variety of information, which presents challenges in storage and retrieval, says Bill Howe, a research scientist with the eScience Institute. “Researchers will need flexible tools for managing heterogeneous data created from a variety of sources,” he says. “This situation is likely to occur in many labs and would be best served if there was a shared resource with core competencies in data management.”
The need for advanced computational and analytical techniques is growing dramatically. As the research model shifts from periodic sampling to continuous monitoring of systems, the amount of data is exploding. “I have a friend at the Monterey Bay Aquarium Research Institute who used to host a group of researchers every other summer who would collect data for a few months, then bring it home and analyze it for a year and a half,” Lazowska says. “Now, they are receiving continuous streams of data. The scientific opportunities and the analytic challenges are both enormous.”
Many researchers are finding the current situation frustrating. They realize they need to do more with their data. Researchers are starting to examine the use of such resources as Amazon Web Services, which potentially can provide huge amounts of computing power. “The advantage of such services is that you pay only for what you consume,” Lazowska says “It may be especially good for research needs that are ‘spiky,’ — intensive for a relatively short period of time. Once more researchers have experience with this, we need to share the knowledge of how to do it and when it’s appropriate. Our interviews show that, across a broad spectrum of top researchers, shared expertise in computing is a greater need than shared computing hardware.”
As research becomes more multi-institutional the challenges of finding sophisticated environments for collaboration are growing. While most researchers now are using the most basic technologies — email and telephone — for collaboration, the need for real-time collaboration tools is increasing, as is the need for expertise to develop and support such tools.
These interviews will help the UW Technology Research Support Working Group to decide what research computing needs are and how to address them. “One of my biggest concerns,” Lazowska says, “is that the current budget climate and uncertainty could lead to a de-emphasis on technology that supports discovery. State-of-the-art check printing is not what makes us a leading research-intensive university. We need to preserve our leadership in networking — not easy given our geographical location. And we need to develop and diffuse expertise in other areas of modern eScience. That’s the only way UW will remain competitive. We need to be foresighted — always making investments that help us anticipate and prepare for the next big thing.”
