|
Google lays out browser aims
By Chrystia Freeland in St Paul, Minnesota and Richard,Waters in San Francisco
Published: September 4 2008
Google launched its web browser in part to stop its rival Microsoft from "Balkanising" the internet by carving it up in ways that favoured its own services, the internet company's chief executive officer said yesterday.
Speaking in a video interview with the Financial Times at the Republican national convention in St Paul, Minnesota, Eric Schmidt said Google's Chrome browser had been built mainly to create a more secure and stable platform for internet users.
However, he also conceded that "there is a defensive component" to the decision as Google tries to prevent Microsoft from using its dominant Internet Explorer [IE7] browser to outflank it. "Microsoft has a history of favouring its own applications and I can give you 500,000 pages of court testimony . . . and so forth and so on about that," he said.
Invoking the famous competitive battle - Microsoft's defeat of browser-maker Netscape that led to a US antitrust investigation - Mr Schmidt added: "We think that the browser continues to be an important platform; that the browser wars of 10 years ago were right: the browser matters."
The launch of Chrome this week marked an abrupt change of course from Google's earlier position and highlighted the high stakes of this latest version of the browser wars.
"It is true that we actually, and I in particular, have said for a long time that we should not do a browser because it wasn't necessary," Mr Schmidt said. "The thing that changed in the past couple of years . . . is that people started building powerful applications on top of browsers and the browsers that were out there, in particular in Explorer, were not up to the task of running complex applications."
Mr Schmidt hinted at Google's long-term aims of challenging Microsoft's dominant position in corporate software, saying Chrome would be a platform for "powerful industrial apps".
"There is an opportunity for a platform and that platform for running these new applications is something that you can't really do on IE7, and that's the argument," said Mr Schmidt.
Mr Schmidt said Chrome, as an open-source piece of software, could not be used to direct browser traffic to Google's own internet services. Asked whether Microsoft should take a similar approach, he added: "We would love Microsoft to open source their browser, but it's unlikely that they would do so in my lifetime, my children's lifetime, or my grandchildren's lifetime."
.
Internet plumbing
Think globally, route locally
Mar 16th 2011, 11:13 by G.F. | SEATTLE
..TAKE two internet technologies designed to serve up web pages, graphics and videos in a jiffy. On its own, each makes man's lot that little bit easier, and browsing less laborious. But run both at once and the internet seems to grind to a halt. What is going on?
The technologies in question cross different axes of the internet. The first involves content distribution networks (CDNs). CDNs shunt content—mostly images, formatting files, audio and video—between a server and browser more efficiently by reducing the number of legs, or hops, a file must travel across. CDNs cache data in servers all over the world, and use indirect clues from browser requests to feed data from the closest server, with proximity gauged in terms of the internet's peculiar topology. This should, in theory, slash download times to the absolute minimum. Akamai, an American internet firm, pioneered the notion. Dozens of companies now offer CDN services, including Amazon and Level 3. (Level 3 secured the contract to provide CDN services for Netflix videos, which led to a conflict with Comcast over peering arrangements.)
The other internet improvement comes in the form of public domain naming system (DNS) servers. DNS translates the server part of an intelligible internet address, like "www.economist.com" in "http://www.economist.com/blogs/babbage", into the underlying numeric address that identifies the entry point to a content-feeding server (or a gateway to multiple servers)—and back again. A computer has to be instructed which DNS server to ask for the translation. Typically, instructions for setting up an account with a broadband provider include these numbers, which need to be plugged into the home or office network router (connected to a broadband modem) whenever a new connection is being configured. The router in turn relays the DNS server addresses to any computer connected to the network (it also assigns a private address to each computer).
DNS was long regarded by broadband firms as a necessary evil because adequate performance was the only requirement. There was no revenue based on rapid DNS look-ups alone. This meant that customers using some providers were in for sometime long waits after a browser asked a computer to ask a provider's DNS server for information. The delay—known as latency—makes web pages appear to load slowly, even if the page itself (and subsequent pages from the same server) loads in a split-second after the DNS process is complete.
Internet business thrives on excising inefficiencies, so users did not have to wait long for OpenDNS. By offering alternative DNS server numbers the company is able to reduce latency substantially. It can also redirect from, say, misspelled domain names to ones that do exist (.cm to .com, for instance), intercept Google queries made in a location field instead of a browser's search field, filter malicious sites and offer parental and business browsing controls. The firm offers its lookups for free, and makes its money from Google ads and academic and business services. Google itself entered the fray with a free, no-features offering in late 2009. The internet search giant offers a multi-platform tool to test from your computer and network which public lookups perform best.
Herein lies the rub. CDN providers use a DNS server's network address to figure out where a user is located, and provide the closest content cache. Since public DNS systems offer few clues as to the precise location of the user, they often end up misleading a CDN service. Rather than charting the shortest possible path through the web's recesses, the upshot is a convoluted one.
Babbage was downloading an Apple test release of iOS recently for his iPad and iPhone 4. His home network has long relied on OpenDNS. For some reason, the download speed was a sluggish 300 Kbps. A broadband network-speed tester hinted that over 15 Mbps in capacity remained untapped. Babbage changed his DNS settings from OpenDNS to his provider's and restarted the downloads. They instantly jumped to the full 15 Mbps. (Apple uses Akamai's CDN service.)
David Ulevitch, OpenDNS's boss, says there is an easy solution. DNS servers currently pass on requests anonymously, with no clues as to the client computer or network making the request. A relatively small tweak in the way these servers request information allows a CDN server to make an informed decision about which cache to offer up. Specific user-address information is not required, only a broad hint as to where on the provider's network the user's requests originate, preserving privacy. And end-user devices do not need to be reconfigured. OpenDNS is already working with CDN providers on a one-by-one basis to enable this feature. The idea arose from a currently circulating draft internet engineering proposal spearheaded by Google and others, and available in DNS server software. (For a time, internet standards preceded implementation; now the opposite is true.)
Mr Ulevitch declines to cite specific CDNs using the feature, but in Babbage's own tests and statistics posted at sites measuring CDN/DNS performance it appears that Akamai has yet to jump on board. Ironically, Akamai used to route most of Netflix's streamed content; Level 3, which currently hosts Netflix, seems to be making ample use of the new location data.
Google operates both public DNS servers and its own proprietary CDN networks for YouTube, advertising and other content. A spokesman kindly pointed Babbage to a policy statement noting that so far as it is concerned, "users in Europe should be directed to CDN content servers in Europe, users in Asia should be directed to CDN servers in Asia, and users in the eastern, central and western U.S. should be directed to CDN servers in those respective regions." The firm is fiddling with the nitty-gritty of its engineering proposal to make it work with any DNS and CDN providers.
The internet grows in fits and starts, and concurrent bursts of evolutionary improvement can affect the whole niche in unpredictable ways. In this case, at least, selection pressure favoured symbiosis.