Boston University researchers have played a major role in the creation of new software aimed at boosting Internet performance. Network measurement techniques developed primarily by Assistant Professor of Computer Science Mark Crovella and his doctoral student Bob Carter have been incorporated into Net.Medic, a consumer software product released this week by VitalSigns, Inc., of Santa Clara, Calif.
Designed to be used with Web browsers such as Netscape Navigator and Microsoft Internet Explorer, Net.Medic allows users to identify the source of slow response times from Web sites and then recommends solutions or automatically fixes the problem. The novel techniques developed by BU researchers form the basis for some of the software's ability to monitor, isolate, diagnose, and correct performance problems on the Internet.
"A slow or unreliable response rate is one of the biggest problems facing Internet users today," says Prof. Crovella. "This new software will show a user the path along which data is being sent to his or her computer by server and identify which network node -- America Online, for example, or MCI -- is at fault if the path stops transmitting data. It will also inform you if you have misconfigured your system or set up your modem incorrectly.
"Our contribution was to develop tools that enable the properties of a particular path through the Internet to be measured in order to find out what the best possible transfer rate over that path could be. This was the underlying measurement that helped VitalSigns to design a product aimed at informing users of all the possible problems that might be causing poor Internet performance."
Prof. Crovella admits to being "very excited" at the release of the product, which has been designed as a mass market consumer item and will be bundled with other products such as modems. "VitalSigns has solved some of the basic problems presented by today's Internet," he says. "It is gratifying to see techniques developed by our group at Boston University incorporated into a product that is contributing to the solution."
The group from which the research originated is the OCEANS Group (Object Caching Environments for Applications and Network Services), founded in 1994 by Prof. Crovella along with Azer Bestavros and Abdelsalam Heddaya, assistant professors of computer science. The group's purpose is to develop new methods to improve the performance of the Internet and the World Wide Web through measurement, analysis, and redesign. The project undertook the first published study, in 1995, of the effectiveness of "caching" on the Web -- the technique by which a browser saves a copy of a Web page so that a user can 'go back' to it without having to re-transfer the page over the nework. Based on that work, group members have gone on to study the benefits of procedures such as eager document dissemination and speculative prefetching. The group's early study of client use of the Web has been widely cited.
"We started working on the problem of improving Web speed and performance before VitalSigns was actually in existence," says Prof. Crovella. "One of our first research papers described how to measure the underlying capacity of the path between client and server, which is limited by the slowest, or bottleneck, link, and the presence of competing traffic. VitalSigns came across this paper when the company was in its startup phase and looking round for whatever technology was available, and as a result, contacted me."
Jim Goetz, vice president for product management at VitalSigns, says that the research by Prof. Crovella and the OCEANS group developed network measurement technology "particularly well-suited" for Net.Medic. "We are looking forward to continuing to work with them on software that will make the Internet a more effective tool for its millions of users worldwide," he says.
The next research goal is to find a way for data on the Internet to automatically avoid traffic jams, say Prof. Crovella, noting that group member Bob Carter is very close to success in that area.
"It is surprising how few other groups are doing research of this nature, just studying the nuts and bolts of the system, measuring it and understanding its properties," he adds. "I don't think people yet realize that as a computer system, the Internet is fundamentally different from anything that has gone before. It is not just a series of computers connected to one another: the problems posed by its immense scale and the relative lack of knowledge that we have about the system itself are actually very interesting research questions."
More information about the OCEANS group and a list of its scientific publications can be found at http://www.cs.bu.edu/groups/oceans/Home.html.