News Blog

A DJ mixed music on stage.

(Credit: Stephen Shankland/CNET News.com)

OK, all you coders toiling in obscurity, are you wondering how the other half lives--the programmers who live the glam rich Internet application lifestyle, ditching Win32 and C++ for Web-based APIs and Python?

A few hundred of them were to be found at the party this week at the Google I/O conference, and I couldn't resist taking some photos. I've been to a lot of trade show parties, and although this wasn't over the top, it was certainly more lavish than the usual rubber-chicken-and-Heineken affair.

What would a Google party be without Wi-Fi? Plenty of people were typing away during the festivities.

(Credit: Stephen Shankland/CNET News.com)

For the event, Google packed Moscone West's third-floor auditorium with games and food--I overcame my fears of Moscone-induced illness and found the sushi tasty--but the real draw clearly was a concert by the witty New Zealand duo, the Flight of the Conchords.

The party had two flowing chocolate fountains with strawberries and other snacks for dipping.

(Credit: Stephen Shankland/CNET News.com)

As usual, the concert was better if you already knew the words. I hesitate to acknowledge that I first encountered the band on YouTube (which had better audio than Moscone) but failed to notify the authorities of the possibility of copyright infringement.

About 3,000 people attended the show, said Vic Gundotra, Google's vice president of engineering in charge of developer evangelism and open-source software. Gundotra seemed to be having a good time at the party--at least until he heard about the binary encoding typo that meant the conference T-shirts read "Google KO" rather than "Google IO."

The conference was alive and kicking. Many sessions were packed to overflowing. It's not clear if the draw was Google's clout or the hunger for information about building rich Web applications, but interest there was.

For those who couldn't be attend, Google plans to post videos of the sessions on the Google I/O Web site in the next week or two.

Among the presentations I joined, I found the most interesting to be Jeff Dean's on the inner workings of Google's data centers; Marissa Mayer's on Google search, and the several talks on Google Gears software to augment browser abilities.

Forthwith, some pictures.

Jemaine Clement, left, and Bret McKenzie form the musical comedy act called the Flight of the Conchords.

(Credit: Stephen Shankland/CNET News.com)

Bean-bag chairs can offer respite after a packed day of sessions on APIs, the Google Web Toolkit, HTML, OpenSocial, and open-source software.

(Credit: Stephen Shankland/CNET News.com)

A member of the Android phone target market checks e-mail during the party.

(Credit: Stephen Shankland/CNET News.com)

Still life with tchotchkes. Nerd toys such as Lego bricks and Rubik's cubes were on hand at the show.

(Credit: Stephen Shankland/CNET News.com)

A partygoer awaits the Flight of the Conchords show.

(Credit: Stephen Shankland/CNET News.com)

Jemaine Clement of the Flight of the Conchords

(Credit: Stephen Shankland/CNET News.com)

Bret McKenzie of the Flight of the Conchords

(Credit: Stephen Shankland/CNET News.com)

The Flight of the Conchords got a lot of laughs from the crowd.

(Credit: Stephen Shankland/CNET News.com)

Plenty of people lined up to shoot a little pool.

(Credit: Stephen Shankland/CNET News.com)

Air hockey afficionados whacked plastic pucks to their hearts' content.

(Credit: Stephen Shankland/CNET News.com)

The Google I/O colored party lights deviated from the standard Google hues. In the background, Wii bowling.

(Credit: Stephen Shankland/CNET News.com)

The beer flowed freely at the party.

(Credit: Stephen Shankland/CNET News.com)

SAN FRANCISCO--The inner workings of Google just became a little less secret.

The search colossus has shed only occasional light on its data center operations, but on Wednesday, Google fellow Jeff Dean turned a spotlight on some parts of the operation. Speaking to an overflowing crowd at the Google I/O conference here on Wednesday, Dean managed simultaneously to demystify Google a little while also showing just how exotic the company's infrastructure really is.

Google fellow Jeff Dean

(Credit: Stephen Shankland/CNET News.com)

On the one hand, Google uses more-or-less ordinary servers. Processors, hard drives, memory--you know the drill.

On the other hand, Dean seemingly thinks clusters of 1,800 servers are pretty routine, if not exactly ho-hum. And the software company runs on top of that hardware, enabling a sub-half-second response to an ordinary Google search query that involves 700 to 1,000 servers, is another matter altogether.

Google doesn't reveal exactly how many servers it has, but I'd estimate it's easily in the hundreds of thousands. It puts 40 servers in each rack, Dean said, and by one reckoning, Google has 36 data centers across the globe. With 150 racks per data center, that would mean Google has more than 200,000 servers, and I'd guess it's far beyond that and growing every day.

Regardless of the true numbers, it's fascinating what Google has accomplished, in part by largely ignoring much of the conventional computing industry. Where even massive data centers such as the New York Stock Exchange or airline reservation systems use a lot of mainstream servers and software, Google largely builds its own technology.

I'm sure a number of server companies are sour about it, but Google clearly believes its technological destiny is best left in its own hands. Co-founder Larry Page encourages a "healthy disrespect for the impossible" at Google, according to Marissa Mayer, vice president of search products and user experience, in a speech Thursday.

To operate on Google's scale requires the company to treat each machine as expendable. Server makers pride themselves on their high-end machines' ability to withstand failures, but Google prefers to invest its money in fault-tolerant software.

"Our view is it's better to have twice as much hardware that's not as reliable than half as much that's more reliable," Dean said. "You have to provide reliability on a software level. If you're running 10,000 machines, something is going to die every day."

Breaking in is hard to do
Bringing a new cluster online shows just how fallible hardware is, Dean said.

In each cluster's first year, it's typical that 1,000 individual machine failures will occur; thousands of hard drive failures will occur; one power distribution unit will fail, bringing down 500 to 1,000 machines for about 6 hours; 20 racks will fail, each time causing 40 to 80 machines to vanish from the network; 5 racks will "go wonky," with half their network packets missing in action; and the cluster will have to be rewired once, affecting 5 percent of the machines at any given moment over a 2-day span, Dean said. And there's about a 50 percent chance that the cluster will overheat, taking down most of the servers in less than 5 minutes and taking 1 to 2 days to recover.

A look at a custom-made Google rack with 40 servers from a modern data center. Infrastructure guru Jeff Dean showed the snapshot at the Google I/O conference.

(Credit: Stephen Shankland-CNET News.com/Jeff Dean-Google)

While Google uses ordinary hardware components for its servers, it doesn't use conventional packaging. . And, Dean said, the company currently puts a case around each 40-server rack, an in-house design, rather than using the conventional case around each server.

The company has a small number of server configurations, some with a lot of hard drives and some with few, Dean said. And there are some differences at the larger scale, too: "We have heterogeneity across different data centers but not within data centers," he said.

As to the servers themselves, Google likes multicore chips, those with many processing engines on each slice of silicon. Many software companies, accustomed to better performance from ever-faster chip clock speeds, are struggling to adapt to the multicore approach, but it suits Google just fine. The company already had to adapt its technology to an architecture that spanned thousands of computers, so they already have made the jump to parallelism.

"We really, really like multicore machines," Dean said. "To us, multicore machines look like lots of little machines with really good interconnects. They're relatively easy for us to use."

Although Google requires a fast response for search and other services, its parallelism can produce that even if a single sequence of instructions, called a thread, is relatively slow. That's music to the ears of processor designers focusing on multicore and multithreaded models.

"Single-thread performance doesn't matter to us really at all," Dean said. "We have lots of parallelizable problems."

The secret sauce
So how does Google get around all these earthly hardware concerns? With software--and this is where you might think about dusting off your computer science degree.

A Google data center, circa 2000. Note the fan on the floor to cool servers.

(Credit: Stephen Shankland-CNET News.com/Jeff Dean-Google)

Dean described three core elements of Google's software: GFS, the Google File System, BigTable, and the MapReduce algorithm. And although Google helps with a lot of open-source software projects that helped the company get its start, these packages remain proprietary except in general terms.

GFS, at the lowest level of the three, stores data across many servers and runs on almost all machines, Dean said. Some incarnations of GFS are file systems "many petabytes in size"--a petabyte being a million gigabytes. There are more than 200 clusters running GFS, and many of these clusters consist of thousands of machines.

GFS stores each chunk of data, typically 64MB in size, on at least three machines called chunkservers; master servers are responsible for backing up data to a new area if a chunkserver failure occurs. "Machine failures are handled entirely by the GFS system, at least at the storage level," Dean said.

To provide some structure to all that data, Google uses BigTable. Commercial databases from companies such as Oracle and IBM don't cut the mustard here. For one thing, they don't operate the scale Google demands, and if they did, they'd be too expensive, Dean said.

BigTable, which Google began designing in 2004, is used in more than 70 Google projects, including Google Maps, Google Earth, Blogger, Google Print, Orkut, and the core search index. The largest BigTable instance manages about 6 petabytes of data spread across thousands of machines, Dean said.

MapReduce, the first version of which Google wrote in 2003, gives the company a way to actually make something useful of its data. For example, MapReduce can find how many times a particular word appears in Google's search index; a list of the Web pages on which a word appears; and the list of all Web sites that link to a particular Web site.

With MapReduce, Google can build an index that shows which Web pages all have the terms "new," "york," and "restaurants"--relatively quickly. "You need to be able to run across thousands of machines in order for it to complete in a reasonable amount of time," Dean said.

The MapReduce software is increasing use within Google. It ran 29,000 jobs in August 2004 and 2.2 million in September 2007. Over that period, the average time to complete a job has dropped from 634 seconds to 395 seconds, while the output of MapReduce tasks has risen from 193 terabytes to 14,018 terabytes, Dean said.

On any given day, Google runs about 100,000 MapReduce jobs; each occupies about 400 servers and takes about 5 to 10 minutes to finish, Dean said.

That's a basis for some interesting math. Assuming the servers do nothing but MapReduce, that each server works on only one job at a time, and that they work around the clock, that means MapReduce occupies about 139,000 servers if the jobs take 5 minutes each. For 7.5-minute jobs, the number increases to 208,000 servers; if the jobs take 10 minutes, it's 278,000 servers.

My calculations could be off base, but even qualitatively, that's enough computing horsepower to make the mind boggle.

Fault-tolerant software
MapReduce, like GFS, is explicitly designed to sidestep server problems.

"When a machine fails, the master knows what task that machine was assigned and will direct the other machines to take up the map task," Dean said. "You can end up losing 100 map tasks, but can have 100 machines pick up those tasks."

The MapReduce reliability was severely tested once during a maintenance operation on one cluster with 1,800 servers. Workers unplugged groups of 80 machines at a time, during which the other 1,720 machines would pick up the slack. "It ran a little slowly, but it all completed," Dean said.

And in a 2004 presentation, Dean said, one system withstood a failure of 1,600 servers in a 1,800-unit cluster.

Next-generation data center to-do list
So all is going swimmingly at Google, right? Perhaps, but the company isn't satisfied and has a long to-do list.

Most companies are trying to figure out how to move jobs gracefully from one server to another, but Google is a few orders of magnitude above that challenge. It wants to be able to move jobs from one data center to another--automatically, at that.

"We want our next-generation infrastructure to be a system that runs across a large fraction of our machines rather than separate instances," Dean said.

Right now some massive file systems have different names--GFS/Oregon and GFS/Atlanta, for example--but they're meant to be copies of each other. "We want a single namespace," he said.

These are tough challenges indeed considering Google's scale. No doubt many smaller companies look enviously upon them.

SAN FRANCISCO--When it comes to search quality, Google has a split personality.

Google uses a method called split A/B testing to measure exactly what changes it should make to its main search Web site--both to its famously Spartan search box and to the results it produces. With the approach, Google shows different versions of the pages to users and measures how they respond, said Marissa Mayer, vice president of search products and user experience, in a speech at the Google I/O conference here Thursday.

Marissa Mayer, vice president of search products and user experience at Google, speaks at the Google I/O conference.

(Credit: Stephen Shankland/CNET News.com)

For example, Mayer said, the company wanted to find out how many search results to show users--the customary 10, or 20, 25, or 30? When asked directly, users said they'd like more results on a page, but testing showed otherwise.

Specifically, Google found that when the results increased to 30 per page, people searched 20 percent less overall, Mayer said. After much analysis of server logs, the company found it was because it took about twice as long to display the longer results list for the user, and speed matters.

"As Google gets faster, people search more, and as it gets slower, people search less," she said.

The same effect happened with Google Maps. When the company trimmed the 120KB page size down by about 30 percent, the company started getting about 30 percent more map requests. "It was almost proportional. If you make a product faster, you get that back in terms of increased usage," she said.

Split A/B testing also led Google to refine exactly how much white space to pad around its logo and other elements on the search results page. And it changed from the industry practice of a pale blue background behind ads to a pale yellow background. People not only clicked on ads more, they also searched more in general, she said.

The subject clearly is close to Mayer's heart. She's an engineer who also has an interest in the more aesthetic realm of design.

"On the Web in general, (creating sites) is much more a design than an art," she said. "You can find small differences and mathematically learn which is right."

A history of Google's search page
Google's search page, with its abundance of empty white space and its almost boastful "I'm feeling lucky" button, looks downright ordinary today. But it wasn't always the case.

... Read more

SAN FRANCISCO--Google showed off working prototypes Wednesday of new possibilities for its Gears project to goose Web browsers' abilities.

When Google launched Gears a year ago, the company overemphasized one important feature, its ability to make Web applications work even when the browser is disconnected from the Internet, Chris Prince, a lead Gears engineer, said in a talk at the Google I/O conference here Wednesday. The new features, though, head in dramatically different directions: notifications on the desktop of various events, support for location information, better interactions with a computer's file system, and technology to let large file uploads proceed even when hampered by intermittent network connectivity.

Chris Prince, a Google engineer, describes new possibilities for Google's Gears software to improve Web browsers.

(Credit: Stephen Shankland/CNET News.com)

At the same time Google is working on the open-source Gears project, competitor Yahoo has begun similar efforts, announcing its BrowserPlus effort this week. The moves show that the Internet giants are trying to steer the basic fabric of the Internet into more lively directions.

"I think people have realized the browser is kind of broken," Prince said in an interview after his talk. "A lot of us are trying to improve it."

The project initially was called Google Gears, but the search giant removed its name Wednesday in an effort to show it's not just the company's work. MySpace announced it's using Gears for its mail system at the show.

Prince wouldn't commit to any of the new features ever making their way into Gears, but it's clear the company has grand ambitions for what can be done with web applications. "We want to make it so Web applications can be just as powerful as desktop applications by unlocking the capability of the local machine," Prince said.

He demonstrated five Gears prototypes:

• One let a Web page create a shortcut icon on a computer's desktop so people could launch that Web application with a double-click instead of a more laborious process.

• A notification process, which like Yahoo's BrowserPlus feature ties into a computer's general system notification abilities, is a major missing piece in letting Web applications seize a user's attention the way desktop apps can. "Web apps have this problem where they can't tell users about important things happening on their system," Prince said.

• His file system demonstration showed a dialog box that let him select a large group of photos for upload rather than the one-file-at-a-time process that today afflicts Web site operations.

• A "blob"-processing ability could be used, for example, to divide a large file into bite-sized pieces, an approach that makes it easier to restore an upload interrupted by a bad network connection.

• He used a geolocation-processing ability to process latitude-longitude information to provide a more useful Google map showing bars near Moscone Center in San Francisco.

Gears still needs to handle privacy, though, when it comes to sharing location information with Web sites, he added. "There has to be permission for using location data. We haven't figured out the best model yet," Prince said.

And though he didn't demonstrate anything, Prince also said there's work under way to try to build Webcam and microphone support into Gears.

A year ago it was Google with its Gears project. Now Yahoo wants to make your browser better, too.

Yahoo BrowserPlus makes it possible to tag, crop, and perform other sophisticated operations for a Flickr photo-uploading Web site.

(Credit: Yahoo)

A year after the Google launched its Gears project, Yahoo announced software called BrowserPlus that has a similar philosophy: expand what's possible to make Web applications a better alternative to programs running natively on a personal computer. Right now, it's available only in a "sneak peek" on some Yahoo-operated Web sites.

"BrowserPlus is a technology designed to 'extend the Web,' so that developers can build more exciting Web applications and so end users can get more done inside their Web browsers," Yahoo said on a BrowserPlus frequently-asked-questions page.

Among its abilities: "Different Web sites can use BrowserPlus to support things like drag and drop from the desktop, easier file uploads, more efficient and secure acquisition of feeds and information, and native desktop notifications," Yahoo said.

This sounds good to me, at least in principle. I've been trying Google's and Facebook's Web-based instant-messaging applications, but without desktop notifications, they're only as immediate and useful as any old Web mail software.

Yahoo's BrowserPlus project aims to make Web browsers more powerful.

(Credit: Yahoo)

The software, along with Gears, shows an interesting trend in Web design: the biggest players are working to expand what can be done with the Internet. It's reminiscent of the early days of the Web, when Netscape and Internet Explorer would implement new features to permit more elaborate Web sites.

Google appears to be trying to make it easier for competitors to embrace Gears. At its Google I/O conference here Wednesday, Google de-branded Gears today, taking its name off the project and announcing some new Web browser support in the works.

BrowserPlus works on Mac OS X 10.4 and 10.5 machines and on Windows XP and Vista machines. Supported browsers are Internet Explorer 7 or later, Apple Safari 3 or later, and Firefox 2 or later, Yahoo said.

Those who want to try it out can visit Yahoo's BrowserPlus demo site. That site offers an in-page Flickr photo uploader that lets users drag, tag, rotate, and crop photos, an IRC IM client; and for the programmer types, a JSON (JavaScript Object Notation) inspector.

SAN FRANCISCO--MySpace said it's using Google's Gears, software for augmenting Web browser abilities, to make the social-networking site easier to use.

When MySpace users go to their mailbox, they'll be invited to install Google Gears, said Allen Hurff, MySpace's senior vice president of engineering, in an appearance here Wednesday at the Google I/O conference. "It's available to everyone today," Hurff said.

Allen Hurff, MySpace's senior vice president of engineering, speaks at Google I/O.

(Credit: Stephen Shankland/CNET News.com)

When users install Gears, they'll be able to quickly search their in-boxes for specific terms or sort messages, for example to show unread mail, Hurff said.

Gears, an open-source plug-in, endows browsers with a number of useful features to make them a better foundation for running elaborate software. Gears hasn't caught on widely, but MySpace gives the project more clout. And Google thinks some of Gears' success is actually measured in its influence over the new HTML 5 standard for describing Web pages.

In MySpace's case, one useful Gears feature is local storage on a computer. Another is the ability to run multiple threads at the same time--in effect, to walk and chew gum at the same time. For MySpace, that includes letting a computer index text at the same time it draws user-interface elements on the browser screen, Hurff said.

On Wednesday, Google announced it's changing the project's name from Google Gears to just Gears.

"We want to make it clear that Gears isn't just a Google thing," said Chris Prince, a Google software engineer, in a blog posting. "We see Gears as a way for everyone to get involved with upgrading the Web platform."

Google also announced that it's expanding browser support for Gears.

"We are currently adding Firefox 3 and Safari support. And Opera is working to support Gears on both desktop and mobile," Prince said.

SAN FRANCISCO--Google demonstrated some new tricks of its Android mobile phone software, including an elaborate use of Google Maps Street View and a touch-screen interface with abilities known for their presence on Apple's iPhone.

Google Street View panned left and right as the Android phone user turned around. Click on the image above for an Android slide show.

(Credit: Stephen Shankland/CNET News.com)

Steve Horowitz, Android's engineering director, used flicking gestures to sweep from the phone's home screen to another during a speech here Wednesday at the Google I/O conference. More unusual, though was a demonstration of how the phone's internal compass and accelerometer can enliven Street View.

After calling up a view of San Francisco using a Web browser, Horowitz turned around, and the Street View screen panned left or right accordingly, reflecting his orientation.

Also new were demonstrations of a central notification service that can display new e-mail, missed phone calls, and calendar appointments; the ability to unlock the phone using a specific connect-the-dots swipe across the screen; an option to put browser or contact list shortcuts on the Android desktop; and a version of Pac-Man from Namco.

Android consists of a Linux kernel with Java virtual machine technology on top for running software. Google supplies many applications, but it's trying to encourage developers to write their own. Google hopes Android will become an open system on which users can install whatever software they want, though it's not yet clear if phone service carriers will agree with that vision.

Although Android supported the touch screen, there was no support yet for multitouch, which permits two-finger controls such as pinching to shrink a photo. However, Android could accommodate that technology if handset makers use multitouch-capable screens, said Andy Rubin, the Android project leader, in a press meeting after the speech.

"When a hardware developer puts that hardware into the handset, I hope that hardware developer provides the driver," Rubin said.

A view of Google's Android mobile-phone software.

(Credit: Stephen Shankland/CNET News.com)

Android can use a touch screen, but doesn't need one, Rubin added. "Steve could have given that entire demo driven by a trackball," Rubin said.

Rubin wouldn't be pinned down about when Android phones will ship, only reiterating the commitment to meet a deadline of the second half of 2008. "What you saw onstage looks pretty good, but we want to make sure it's perfect," Rubin said.

In the demo, Android ran on a UMTS (Universal Mobile Telecommunications System) phone from an unnamed manufacturer, Rubin said. It used a Qualcomm MDM 7201A processor, a Synaptics capacitive touch screen, and a 3.6 megabit-per-second HSDPA (High-Speed Downlink Packet Access) broadband connection.

Google is expected to update its Google Web Toolkit (GWT) this week at its new developer conference, according to eWeek.

GWT is designed to help programmers write richer Internet applications using a beefed-up JavaScript programming technique called Ajax; the project was released as open-source software in 2006 with version 1.3, and the current version is 1.4. There are several GWT talks at the Google I/O conference.

Google has been working on improving GWT's performance, Java compatibility, and developer tools, eWeek said.