Google's SPDY accelerator gets new wind in its sails
The technology for speeding up Web pages becomes part of the standard foundation of the Web. But should that standardization effort reach beyond just SPDY?
- Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
Has a slow Web been getting you down lately?
Just imagine if your multibillion-dollar business depended on it, as Google's does. Then imagine the glee in Google's corridors at a significant new victory in the company's attempt to build a Web-accelerating technology it calls SPDY into the Internet.
Earlier today, Mark Nottingham, chairman of the HTTPbis Working Group, announced support for SPDY in an overhaul of one of the networking foundations of the World Wide Web. That foundation is HTTP, the Hypertext Transfer Protocol, and Google hopes SPDY will open up some of its bottlenecks.
The HTTPbis group, part of the Internet Engineering Task Force (IETF), has been been sprucing up the 1999-era HTTP 1.1 for several years. But Nottingham said it's now time to look to the future.
"There seems to be broad agreement that the time is ripe to start work on a new version of HTTP in the IETF, and that it should happen in this Working Group," Nottingham said. When work refurbishing HTTP 1.1 began, there wasn't interest in a new version of HTTP, but SPDY and its adoption shows there's interest now, he said, proposing completion of a draft of HTTP 2.0 by May and finish the work by July 2013.
Speed boost
In Google's research, SPDY accelerates page-load times by 28 percent to 43 percent over a 2Mbps DSL line and 44 percent to 55 percent over a 4Mbps cable broadband connection.
Standardization is often a drawn-out, painstaking affair, but it can pay off by making a particular technology easier to incorporate in a wide range of products. A neutral industry standard can be technically, politically, and legally easier to embrace.
Tinkering with the basic workings of the Web is a tricky business given the immense variety of browsers, servers, and network gear in between. Google has made progress with SPDY, though, since unveiling SPDY in 2009 and then building it into its Chrome browser.
SPDY is a high-profile element of Google's "make the Web faster" effort. Yesterday, the company also detailed proposed improvements to TCP an even more fundamental Internet technology. The Transmission Control Protocol governs how a huge amount of data is sent over the Net despite problems such as network congestion or lost packets of data.
Google is in many ways perfectly positioned to rebuild the Net. It's got the world's most-trafficked Web sites and measurements that show that speed means profits. It's got the No. 3 browser, Chrome, so it can experiment with technology that works hand-in-hand with its Web sites. And it's got an army of research-minded engineers encouraged to think big.
Those assets don't always equate to success, though, when it comes to rewiring the Web--in particular given that Google's self-interest means that its ideas can raise competitors' hackles. Microsoft and Mozilla, for example, while sharing Google's interest in making the Web a better foundation for programming, are leery of Google's Dart programming language that competes with the incumbent JavaScript.
SPDY milestones
One earlier milestone was SPDY's inclusion in Amazon's Silk, the browser on its Kindle Fire tablets. A more recent one is that Firefox 11 introduces SPDY support, too.
"I think with Firfox on board with SPDY, it's got legs," said Mike Belshe, who along with Roberto Peon invented SPDY at Google. (Belshe since has moved to a start-up, Twist. "We'll get it (or its derivative) standardized in 2012 for sure."
Technical details and a SPDY white paper show how it works and what it offers. It employs a number of tricks to speed up Web page transfers, including compression, prioritization of important Web page elements, and a way to sidestep today's limits on opening multiple network connections.
One technical detail that historians might be interested in is what exactly SPDY stands for. The answer: nothing. "It doesn't stand for anything," Belshe said, "but it sounds like 'speedy.'"
Adding SPDY isn't the only possible way to improve HTTP, and Nottingham acknowledged that it will be a "tightrope walk" to admit some HTTP improvements without getting bogged down in massive rework of the technology.
FreeBSD programmer Poul-Henning Kamp was unenthusiastic about Nottingham's proposal, though, because he sees it as too narrow in scope. He also asked where in Nottingham's proposal there is room for evaluating other suggestions
"In my mind, the effort [Nottingham] sketched out would be correctly titled 'Beatify the SPDY protocol as a carrier of HTTP/1.1 traffic,'" Kamp said in his "devil's advocate" response. "HTTP/2.0 would in my mind be an attempt to actually improve the protocol."