An MIT Boost for the Web

Polaris Tool Loads Pages 34 Percent Faster

Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory and Harvard University have developed a piece of software, dubbed Polaris, which they say can reduce page load times by as much as 34 percent. The researchers expect to present their findings at the USENIX Symposium on Networked Systems Design and Implementation, held this week in Santa Clara, California.

According to the Polaris team, “It can take up to 100 milliseconds each time a browser has to cross a mobile network to fetch a piece of data”. “As pages increase in complexity, they often require multiple trips that create delays that really add up. Our approach minimizes the number of round trips so that we can substantially speed up a page’s load time.”

When Web pages are requested, browsers reach across the network to deliver objects. The browser then evaluates and adds the object’s content to the page that the user sees. But the process, though seamless, is not currently optimised for speed. Browsers do not see all of a page’s dependencies, owing to the way HTML is structured, resulting in an increase in the number of network trips and slow page loads.

Polaris tracks all of the interactions between objects. For example, when one object reads the data in another object, or updates a value in another object, it uses a detailed log of these interactions to create a dependency graph for that page, which Polaris then uses to optimize how Web pages are loaded.

Evaluating one object often means having to fetch and evaluate more objects, which are described as dependencies of the originals. For example, a browser might have to execute a page’s JavaScript code in order to discover more images to fetch and display.

Simple pages are not expected to benefit much from the tool, but more-complex pages — which are the norm these days — should see shorter load times. The researchers say they’ve tested Polaris on a range of network conditions on 200 of the top-ranked Web sites.

Google and Amazon have also tried to reduce page load times, but by using data compression. Proxies like Opera Mini use an allied technique to squash Web pages for less-powerful browsers on mobile devices.

The researchers plan to “eventually” open source Polaris, but in the near term they hope to convince browser developers to embed their technology. Polaris is no panacea for slow browsing speeds, but complex sites with a vast number of dependencies stand to particularly benefit. Mobile networks could also see improvements as the tool optimises the number of fetches on these networks.