Monday, July 7, 2008

ISPs prepare for video revolution

Video may have killed the radio star, but it doesn't have to kill the Internet.

That is if Internet service providers can figure out how to keep up with the video-driven bandwidth demand on their networks. Peer-to-peer technology provider BitTorrent says it can help.

Video consumes more network resources than any other media distributed on the Web. Even poor-quality video from YouTube eats up more bandwidth than e-mail, music downloading, and voice over IP services. And when you throw full-length high-definition video into the mix, you're talking about even more bandwidth. Depending on the compression used, a single HD video stream can eat up 20 megabits per second worth of bandwidth.

And as consumers subscribe to faster and faster broadband connections at home and sites like YouTube and Hulu come online offering all kinds of video choices, more people are watching video on the Web. According to ComScore Video Metrix, Americans are currently watching upward of 10 billion videos online a month. By the end of 2007, online viewers averaged more than one video a day.

This is just the beginning. ABI research forecasts the number of viewers who access video via the Web will nearly quadruple in the next few years, reaching at least 1 billion in 2013.

This summer's Olympic Games in Beijing marks the first real test of online video as NBC embarks upon the most ambitious online video project ever. NBC plans to offer 3,600 hours of live programming from Beijing, which translates to about 212 live hours for each of the 17 days of the Olympics. The majority of this viewing will be delivered online.

All this video is great for viewers, who are able to pick and choose what they watch and when. But for Internet service providers like the phone companies and the cable operators, it represents a massive challenge. Some providers, such as Comcast and Time Warner Cable, are testing out new ways to deal with "bandwidth hogs" or individual users who use an inordinate amount of bandwidth.

Last month, Comcast began testing a new system that will throttle back or slow down traffic during times of congestion for heavy bandwidth users. The new system was developed after Comcast faced stark criticism for singling out and slowing down peer-to-peer traffic.

Meanwhile, Time Warner Cable, which says it faces the same capacity headaches, also began testing a new billing system that charges customers who exceed their limit for uploading and downloading material.

P2P as a solution, rather than a problem
Eric Klinker, chief technology officer for BitTorrent, which has commercialized the peer-to-peer technology, says that what the cable operators are doing is a good start. But more can be done to help operators deal with the onslaught of video.

"I think what Comcast and Time Warner Cable are doing is a great first step. It gets ISPs out of the business of deciding which applications are important and which aren't. But there are enhancements to the peer-to-peer protocol, in particular, that can make it easier on all ISPs."
--Eric Klinker, CTO, BitTorrent,

For one, peer-to-peer protocols, such as BitTorrent, which are often cited as major headaches for network operators because of the big file transfers they enable, need to be utilized rather than singled out as a source of the problem, he said.

"I think what Comcast and Time Warner Cable are doing is a great first step," Klinker said. "It gets ISPs out of the business of deciding which applications are important and which aren't. But there are enhancements to the peer-to-peer protocol, in particular, that can make it easier on all ISPs."

Peer-to-peer technology has gotten a bad rap for years. Since the days of file-sharing networks like Napster, which allowed people to exchange songs on their computer hard drives with others on the Internet, peer-to-peer technology has been demonized in the press. But the truth is that peer-to-peer technology actually allows large files like videos to be distributed more efficiently. And as more video makes it way onto the Web, it's increasingly being used. In fact, peer-to-peer traffic accounts for about 43 percent of all traffic on the Internet, according to a recent study by the network management company Sandvine.

The way peer-to-peer works is that when a user requests a video, the peer-to-peer network queries other users in the network and takes pieces of the file from different peers and sends it to the user requesting the file. This distributed architecture means that content owners don't have to assemble large and expensive data centers. It also means that a content distributor doesn't have to pay for expensive high-speed links to serve up an entire file from a single server farm.

That said, peer-to-peer protocols in the wild can eat up lots of bandwidth because peers on the network can silently and continuously upload pieces of files from their computers all day and all night, seeding dozens or hundreds of file requests. And because upload capacities are generally much slower than downloads, it can create bottlenecks and capacity crunches on the last mile of service providers' networks. For network operators that are already capacity-constrained, this phenomenon can dramatically affect performance for all users.

The enhanced version of peer-to-peer
Peer-to-peer companies, such as BitTorrent and Pando Networks, have recognized this problem and have been working with service providers, such as Verizon Communications, Comcast, and others to come up with solutions. Verizon and Pando Networks have been working on a project called P4P, which advocates ISPs share information about their network topography and use an enhanced version of peer-to-peer to locate peers in close proximity to the file request. Getting files locally can help reduce the expense associated with carrying peer-to-peer files over long distances.

BitTorrent, whose founder created one of the most popular peer-to-peer protocols used today, has also been working on a solution. The company has developed its own enhancement to the peer-to-peer protocol that tells peer-to-peer applications to stop seeding the network with content when the network is congested.

For example, if a teenager starts playing an online video game at the same time his mother makes a voice over IP phone call and his little sister is downloading music from iTunes, the protocol will tell the peer-to-peer movie application that is running in the background on their family computer to stop uploading bits of the Spider-Man movie that had been ordered from an online movie rental service and is now stored on a hard drive in their home. Instead, the network will search for the content on another peer that is on a network that is less congested.

"If there is contention in the network, my application will back off," Klinker said. "And it will automatically and seamlessly find someone else in the network to complete uploading that content. The video quality is never disrupted, and the user never knows where the content is coming from."

BitTorrent has already tested the enhancement with more than 10 million users and it's currently working with the Internet standards body, the Internet Engineering Task Force, to standardize the technology so that other peer-to-peer companies can embed it in their software client.

Klinker said that new technologies, such as the one developed by his company, as well as efforts on the P4P enhancements will help ISPs manage and control their networks so that even more video can make it to the Web without crippling the infrastructure delivering it.

And once service providers learn how to harness peer-to-peer, they will be able to develop business models that reap the benefits of the technology. For example, Comcast, Verizon, or any other TV provider could add peer-to-peer software to the set-top boxes sitting in their customers' living rooms to create a distributed peer-to-peer movie network. Instead of serving up on-demand movies from their own servers sitting in expensive-to-run data centers, these TV providers could leverage the content already stored on their customers' set-top hard drives.

"The cable and phone companies are already spending capital to put set-tops in everyone's home," Klinker said. "They could use that same hardware as part of their content distribution model. Then the user pays the electrical bill. And they pay for the bandwidth. It's just much more efficient."

Klinker said a solution, such as this one, requires a slightly different business model from BitTorrent's current business model. But he said that it's something the company is investigating.

"It's interesting enough that we're in discussions and testing some scenarios with ISPs," he said. "In general, service providers move slowly. So nothing will happen overnight. But I think we'll see some interesting changes within the next three years."

Apple MacBook: Change is in the Air

Posted by Brooke Crothers

The Apple MacBook Air has been a ground-breaking first-generation product (in my opinion). So, what will Apple do to top it when an update comes later this year? There are some telling indicators already. This is what I expect--and hope for--as a user.

(Credit: Apple)

First, a disclaimer. I am not an Apple fanatic. The MacBook Air is the first Apple product I have ever used for more than a few days. For well over a decade, I have been wedded to Wintel (Windows-Intel) laptops.

Before I dive into upcoming features, I should also mention that I have been extremely pleased with the Air and have used it almost daily for the last four months. But I would be remiss if I didn't say it is overpriced, as all subnotebooks are.

• Overpriced but still an amazing design Apple made a very studied decision to exclude certain features. This makes the Air an Air. Apple could have included more ports and a little more of this and pinch of that--but then it would have been just another subnotebook.

So, I expect Apple to maintain the uniqueness of the Air for the next refresh.

But improvements are always welcome. And here are a few things that potential buyers can expect to see when a new Air is rolled out.

Apple has begun to give us hints of things to come. A $500 price cut for the solid state drive (SSD) model is one of the biggest indicators so far.

• A bigger, better solid state drive The next Air will offer drives that range in size to more than 100GB. A likely offering would be 128GB from vendors like STEC. (Samsung supplies the current SSD.) Intel and Micron Technology can't be ruled out. Their drives will come in 80GB and 160GB capacities.

These SSDs will also likely use multiple-level cell (MLC) technology, in contrast with current drives that use single-level-cell (SLC). MLC allows higher-capacities but presents power and data reliability challenges, which suppliers claim to have overcome.

• Processors Invariably, all notebooks get upgraded with better processors and graphics. I think the Air's current performance is superb for a subnotebook. I have owned many subnotebooks over the years and anemic performance can render them practically unusable as an everyday machine. But I haven't had this problem with the Air (see note at bottom).

Intel's upcoming 45-nanometer "Montevina" (Centrino 2) low-power offerings should make this experience even better. Though an initial Montevina refresh is slated for July 14, low-power versions won't appear until this fall. Intel refers to these as SFF (small form factor) processors. They will come in high-performance, low-voltage, and ultra-low-voltage variants.

SFF Montevina processors will range from 25-watt (2.4GHz) to 17-watt (1.86GHz) to 10-watt (1.2GHz). The current Intel processor used in the Air is rated at 20 watts at 1.8GHz.

Whether Apple chooses one of these or opts for something not currently on the Intel roadmap of course remains to be seen.

• Graphics Graphics will get upgraded. Montevina will come with Intel's GMA X4500 graphics, which Intel has said repeatedly will be three times faster than current X3100 integrated graphics.

Battery Insufficient battery life is a problem that plagues all subnotebooks. It has often been suggested that Apple include a removable battery (for easy replacement), but that could compromise the ultraslim design. Having said that, I have been pleased with the battery life compared with other notebooks I have owned.

Hazarding a guess at other features such as upgraded hard disk drives, better screens, and external extras like a docking station is too speculative (and the latter would also compromise the design), so I'll refrain from making any predictions.

But the Air shouldn't change too much. With a simple performance upgrade, it would be an even more remarkable computer.

(Note: No, the Air is not as fast as a 14-inch Hewlett-Packard 6910P, for example, but no PC maker can squeeze that kind of performance into a Air-like form factor.)

LiMo gets Openwave browser and messaging


Published: July 7, 2008 6:15 AM PDT


Purple Labs, an increasingly prominent mobile Linux firm and a member of the LiMo Foundation, has bought the browser and messaging side of Openwave's business.

Purple Labs joined the LiMo Foundation--one of three big, mobile open-source consortia, alongside the Symbian Foundation and Google's Open Handset Alliance--at the start of 2008. Working alongside NXP Semiconductors, it gained further exposure at the end of January by launching the Purple Magic phone, a cheap, Linux-powered, 3G handset that could be sold to operators for less than $100.

Openwave supplies mobile-client software to 12 of the top 20 global mobile operators, including Orange, Telefonica, T-Mobile, Virgin Mobile, Vodafone, Sprint, and Verizon Wireless. It also supplies fixed-line broadband providers with software. Around 80 employees are expected to migrate to Purple Labs, including the general manager of the business unit, Gordon Tsang.

Through the $32 million acquisition of Openwave's mobile-client business, Purple Labs said it now supplies mobile-browser software to all of the top five phone manufacturers, which together produce over 80 percent of the world's mobile phones.

Purple Labs' chief executive, Simon Wilkinson, sent a letter to handset manufacturers on Thursday, in which he said Openwave's existing products would be maintained. He also said Purple Labs would "invest in the next-generation Surfer browser and other advanced mobile-internet technologies."

"Further, we plan to reuse many of the Openwave technologies in our Purple Labs Linux platform," wrote Wilkinson. "We are focused exclusively on the requirements of LiMo Foundation members, such as Vodafone and Orange, and believe that this will ultimately create new market opportunities for our manufacturer customers."

In a statement earlier this week, Openwave's chief executive, Bruce Coleman, said his company would now focus on "network-based offerings."

David Meyer of ZDNet UK reported from London.


Saturday, July 5, 2008

Stolen: Google employees' personal data