Taking just two months to dash through a project once anticipated to need a year or more of work, a team at Johns Hopkins has created a way to search three different astronomy databases seamlessly and simultaneously via the World Wide Web.
The new service, known as SkyQuery, works through a programming approach known as Web services, a revolutionary new technique that allows various kinds of computers in a variety of locations across the Internet to cooperate in providing a service to users.
SkyQuery is an important step forward for the National Virtual Observatory, a project supported by the National Science Foundation and currently in development at Hopkins and many other research institutions. NVO scientists want to help astronomers take full advantage of the massive amounts of astronomical information piling up in databases worldwide. Astronomers have begun to pull in data at such an incredible rate that there's great potential to make important new discoveries in the backlog of data without ever looking up at the sky. NVO scientists plan to use the latest computer technology and data storage and analysis techniques to establish a unified and relatively painless way to access as many databases as possible through one online portal.
"SkyQuery demonstrates this principle by making access to three large astronomy databases entirely transparent," says Alex Szalay, co-principal investigator of the NVO and Alumni Centennial Professor of Astronomy in the Krieger School of Arts and Sciences. "You can run a query as if these three databases, located at sites across the country, were a single database."
Szalay, who has a joint appointment in Computer Science, says he and other NVO investigators had once thought that getting a first, prototypical search engine online for NVO might take up to three years. SkyQuery came together quickly both as a result of the design team's intense efforts and as a result of a Microsoft programming contest, the Microsoft.Net Best Awards for 2001 for XML WebServices.
SkyQuery won second prize in that contest, which challenged college students across the country to write new programs that took advantage of .NET, a new Microsoft Web services platform.
Microsoft awarded contest prizes at a reception on April 10 in New Orleans at TechEd 2002, a Microsoft-sponsored conference for software developers and other information technology professionals. As second-prize winners, Johns Hopkins team members will receive $10,000, and Microsoft will donate an additional $10,000 to a Johns Hopkins scholarship fund.
After an initial call for proposals for the contest, judges from the computer industry selected 100 finalists in January 2002. Like the other contestants, the Hopkins team, comprised of Szalay; Tanu Malik, a Computer Science graduate student; Tamas Budavari, a Physics and Astronomy postdoctoral fellow; and Ani Thakar, an associate research scientist in Physics and Astronomy, had to have their finished program to Microsoft by March 15, little more than two months later.
"Nothing like a little pressure," jokes Szalay, who adds that the bulk of the responsibility for getting the program put together in time fell on Malik, who served as the team's leader, and Budavari. "I have rarely seen people that exhausted."
Recalls Malik, "All of us went to a conference at Microsoft, to learn what .NET is all about. When we came back from that conference, then we started implementing it, and it was like three weeks, day-in and day-out, working on it."
Thakar credited the extraordinary capabilities of Microsoft's .NET platform with making the tight deadline doable.
".NET made the job about 10 times easier than it would've been otherwise," Thakar says. "It supplies the framework and all the building blocks necessary to develop Web services in a standardized and easy way."
The team's finished product can be accessed at contest.eraserver.com/SkyQuery.
With an understanding of the language used to run database queries, any visitor to the page can run a search that simultaneously accesses data from the Sloan Digital Sky Survey, or SDSS, in Batavia, Ill.; the Faint Images of the Radio Sky at Twenty-centimeters survey in Baltimore; and the 2 Micron All-Sky Survey, or 2MASS, in Pasadena, Calif. Both the SDSS and 2MASS are currently expanding with new data that will be accessible through SkyQuery.
One of the bigger challenges of SkyQuery, according to Budavari, was developing the algorithms that the service uses to determine when data in different databases come from the same astronomical object.
"We developed ways to use behind-the-scenes info in all three databases to help us do probabilistic matches and figure out which objects are the same," Budavari explains.
Budavari adds that SkyQuery is "nonstatic." Other databases can and will be added in the future to increase the power of the search engine. He did, however, have one caveat about SkyQuery. It seems that in the rush to get things together for the Microsoft competition, one element of the SkyQuery portal slipped behind.
"We're going to be working to make the Web portal a little more aesthetically pleasing," Budavari says with an embarrassed smile.
Other top finishers in the Microsoft contest include a team from the University of Minnesota, Twin Cities, that won first prize for Renderfarm.NET, a program that transforms three-dimensional data into animations and images; and a team from the University of Bridgeport that won third prize for a program called Brain Webber QA that lets users track and manage defects in online endeavors and other projects.