Researchers at MIT say a framework they've developed for browsers can reduce page load-times by up to 34 percent and enhance data-compression techniques used in Chrome.
MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) and Harvard University researchers argue that the framework could, if adopted by browsers, make a significant contribution to existing compression-focused efforts to cut page load-times.
The researchers believe their framework can contribute to the ongoing quest for faster mobile-page load-times, highlighting that fickle mobile users will abandon pages over millisecond delays.
"It can take up to 100 milliseconds each time a browser has to cross a mobile network to fetch a piece of data," said PhD student Ravi Netravali, a contributor to the Polaris paper.
"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," he added.
The researchers compare the system to providing a traveller with a list of cities ahead of a trip so the individual can plan the fastest route.
The paper will be presented at this week's USENIX Symposium on Networked Systems Design and Implementation, where Google researchers last year revealed its Flywheel proxy compression technology for Chrome.
FlyWheel is also focused on improving the web for mobile-device usage. However, it was aimed at reducing mobile data consumption and related costs rather than load-times.
FlyWheel compressed web content by about 50 percent. But while it cut load-times for large pages, it caused higher load-times for small pages.
The researchers behind Polaris say they tested the system under a range of network conditions on 200 popular websites, including Apple.com, ESPN.com, NYTimes.com, and weather.com.
The reduction of up to 34 percent was measured on tests "run with latencies ranging from 25ms to 500ms, and bandwidths ranging from 1Mbps to 25Mbps".
The largest load-time reductions were achieved on ESPN and weather.com, which had the most complex dependency graphs, while for Apple.com Polaris made little difference.
They also found the benefits of Polaris increase as network latency increases, because "higher RTTs (round-trip times) increase the penalty for bad fetch schedules". So, they argue, Polaris could be particularly useful for poor networks.
However, they note that all tests were run on "cold browser caches" and that Polaris offers little advantage over a standard scheduling algorithm on sites, for example, like eBay, which caches 92 percent of all objects.
Still, the CSAIL team claims Polaris' improvements on load-time "are more consistent and more substantive" than compression techniques.
Professor Hari Balakrishnan, another contributor to the paper, commented: "Rather than decreasing the number of transferred bytes, we think that reducing the effect of network delays will lead to the most significant speedups."