A team of academic researchers have developed a pair of tools which combine to improve web browser speeds by over 30 percent, according to a research paper released this week.
Ravi Netravali, Ameesh Goyal, and Hari Balakrishnan from the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT and James Mickens of Harvard University will present the findings published in “Polaris: Faster Page Loads Using Fine-grained Dependency Tracking” (PDF) at the USENIX Symposium on Networked System Design and Implimentation next week.
Browsers traditionally gather the elements of a page in the order determined by a dependency graph, but hidden dependencies force them to make conservative assumptions about that order, making inefficient use of network and CPU resources. The researchers addressed this inefficiency by developing “Scout” to capture edge dependencies by tracking fine-grained data flows, and dynamic client-side scheduler “Polaris” to load objects in an order determined by fine-grained dependency graphs.
“The objects in a web page can interact in complex and subtle ways; however, those subtle interactions are only partially captured by lexical relationships between HTML tags,” the researchers said in the report. “Unfortunately, prior load schedulers have used those lexical relationships to extract dependency graphs. The resulting graphs are underspecified and omit important edges.”
The researchers explain the conservatism browsers are forced into by partially revealed, coarse-grained dependency graphs with the example of browser evaluation of first.js tag, which may or may not modify downstream objects.
The researchers also told Mashable that although it has yet to be tested with various browsers, Polaris is in principle browser-agnostic, and also that there is no increased risk of loading malicious code or malware associated with it.
Most of the effort into improving the speed of websites seems to focus on the server side, where there are a number of measures site operators and web hosts can take. Load times for many pages can also be improved with adaptive database engines.
Google discontinued its PageSpeed service last year, leaving the market to CDNs like Cloudflare.