Follow
With MIT’s freshly developed algorithm, Turbonet may be closer than we think.
A new algorithm developed by a team of researchers at the Massachusetts Institute of Technology could cut wait time for web pages to load by a third. Researchers said that the new technology dubbed Polaris helps browsers load complex web pages 34 percent faster than available tech.
The team explained that existing browsers need 100 milliseconds to track each object’s location on a website when they load it. And because sites are now laden with thousands of objects such as images, HTML files, and JavaScript codes, browsers need to perform several trips that make load times painfully long.
Ravi Netravali, lead author of a study on the new tech, noted that Polaris tends to keep the number of trips to a minimum by automatically mapping out connections between objects and their interactions.
The tech is not ready to be launched yet, but it has been tested on 200 complex websites.
MIT researchers said that existing browsers need to fetch multiple objects on websites because they do not know yet what pages a user would access. Plus, there is a problem with the original objects’ ‘dependencies,’ which are short evaluations of those objects. The browser needs to be extra careful with the order it loads those dependencies because they have a HTML form, making them more difficult to locate.
The issue considerably slows down load times because a single page can contain thousands of these objects. But Polaris created a detailed log of the interactions between objects and dependencies to generated a graph for the dependencies on the next pages.
James Mickens, a Harvard researcher who was also involved in the work, likened the process with a business trip. For instance, if a businessman has in mind a list of remote cities to visit, he may find that he doesn’t like visiting some of the cities before returning home.
But if that person had a map of the cities and their points of interest, he would have not be forced to ‘zig zag’ between those cities to find what he was most interested in. He could have planned the trip beforehand and save a lot of time and effort.
“For a Web browser, loading all of a page’s objects is like visiting all of the cities,”
Mickens added.
So, the new algorithm actually creates a ‘map’ with all the cities before you even start your trip. This is how pages will load a lot faster, the team explained.
Image Source: Flickr
