Sun Labs Looks at Search, Online GamesBy Jeffrey Burt | Posted 2006-12-22 Email Print
WEBINAR: Event Date: Tues, December 5, 2017 at 1:00 p.m. ET/10:00 a.m. PT
How Real-World Numbers Make the Case for SSDs in the Data Center REGISTER >
At the Labs' facilities in Massachusetts and California, researchers work on technologies they hope will eventually make it into commercial products.
BURLINGTON, Mass.Sun Microsystems spends about $2 billion a year in research and development, touching on everything from its SPARC and x86 server lines to its Solaris operating system to its storage devices and Java development offering.
About 10 percent of that R&D money goes to the company's labs, where researchers pursue different projects with the goal that parts or all of what they develop will find their way into commercial products, according to Robert Sproull, vice president and Sun Fellow at Sun Labs.
"We and our industry depend on a lot of basic innovation," Sproull said.
The Labs' 150 or so engineers are usually working on several dozen projects at one time, both here and at the other location in Menlo Park, Calif.
In early December, Sproull and other engineers gave several journalists and analysts a peak at some of the projects being worked on by the Labs group, and ways the technology could be put to use in the enterprise.
The projects now are in the prototype stage, and there's no guarantee they will progress beyond that. But like the prototype cars automakers unveil every year, some of this technology most likely will find its way into commercial products.
Two of the projects focus on developing better search engines. Steve Green, principle investigator in Sun Labs' Advanced Search Technologies group, said that about 85 percent of information generated by businesses is unstructured, and that knowledge workers spend about 25 percent of their time looking for information.
The goal of Green's project is to make search technology more effective.
Using Amazon as an example, Green pointed out that, when a customer orders a book, Amazon's search engine tells the customer the titles of other books that other customers have ordered with that same book. The problem, Green said, is that because two books were bought at the same time doesn't mean that they're similar.
The search engine he's working on would use particular key words to link books together, leading to a greater likelihood of a closer match. Applying that same idea to searches of business data in an enterprise could make it easier to find information, he and Sproull said.
"The whole goal of search technology is to reduce the amount of categorizing," Sproull said.
Paul Lamere, another principle investigator, is using that same idea in his Search Inside the Music project. The goal is to make it easier for users to search through the massive number of songs stored on such devices as iPods and cell phones, Lamere said.
The technology finds similarities in songsfrom structure to instrumentsso that a user can more easily cluster similar music together.
The prototype developed by Lamere offers a variety of visuals, one that looks like multi-colored stars floating in space. The different colors represent different musical genres, from rap to country to rock to classical, and users can click on a dot and see the song and album art.
Next Page: Who's at the meeting?
The technology hopefully would solve a number of problems that come with having thousands of songs stored in a device, Lamere said. Twenty-three percent of the songs are played 80 percent of the time, and 64 percent of the songs stored on these devices are never played, he said.
"People just forget they're there," Lamere said.
Having a search engine that can more closely cluster like songs could improve that, he said. Commercial applications for the technology could include selling it to cell phone companies, which could turn around and offer it to users as an extra option to customers who store music on their mobile devices.
Nicole Yankelovich, another principle investigator, is working on a project aimed at improving the effectiveness of distributed meetings, where some people are attending remotely.
"Some of the problems we're trying to solve here are basic conference problems: who's here, who's talking," Yankelovich said.
Sun's Connected Conference Room prototype lets people see such things as who's talking, who's in attendance and who's voting which way. A menu of audio problemsfrom the sound being too soft, or even lost, to distractionsalerts people to issues.
In addition, documents can be shared, access controlled and conversations held outside of earshot of others in the meeting. Text chat is offered and a "waiting room" is offered for people waiting to come into the conference but who aren't allowed into the meeting until their turn.
Another feature they're working on would allow remote users to control a camera so they could scan the meeting to see who is attending. In addition, a monitor in the room would enable people in the room to see the remote worker's face.
Project Darkstar is aimed at improving the process of designing massive online games, a small but growing part of the industry that Karl Habert, a director in Sun Labs, said is expected to grow to $11 billion by 2011.
Such games now are primarily deployed and run by well-financed companies, which have the money and capabilities to handle the massive amount of data center space and resources needed for this space. Habert estimated is costs about $20 million to design a data center to run a massive online game and then develop the game itself.
"Building an online game [infrastructure] is very complex," he said. "There's a high capital outlay."
Darkstar is a Java-based platform designed to take care of the basic technology needs required to run a massive, multiplayer online gamefrom the load balancing and communications to database performance tuning and scalabilityfreeing game designers to focus on their applications.
People interested in creating such an online game could use the Sun technology and use it to create their infrastructure, and put their efforts into creating the games.
"We want to really open up the MMOG [massively multiplayer online game] space," Habert said.
It also will enable users to access the game from multiple devices, whether it be a home computer or a cell phone.
Sun, of Santa Clara, Calif., is looking at business models for the technology, including whether to lease it to third parties or host it internally. "Sun hasn't committed to building a service around this, but we're thinking about it," he said.
The prototype is currently running in Sun Labs, and is available in "early access" form, Habert said.
Check out eWEEK.com's for the latest news, views and analysis on servers, switches and networking protocols for the enterprise and small businesses.