A Better Way to Stream TV?

In a model that could galvanize the Net, Israeli startup Arootz says it can satisfy Web video demand using multicasting and cheap hard-drives

Watching YouTube clips, TV shows, and even full-length movies over the Internet seems like a great idea. There's lots of choice, you can watch when you want, and you can send links to your friends. Plus, the great global network has virtually unlimited bandwidth, right?

Wrong. The demands of streaming video over the Net are already starting to put a strain on both the infrastructure and the business models of the world's Internet service providers (ISPs). Networks built to accommodate e-mail, conventional Web surfing, and occasional downloads are buckling under the challenge of serving up millions of individual, high-speed data streams to multimedia-hungry customers.

Meanwhile, when data consumption was lower, ISPs could sell essentially the same bandwidth to 50 customers for about $30 per month and call it "unlimited" access. Most of the time, the collective usage of those subscribers averaged out so that nobody knew they were sharing the same wire with dozens of other people. But when video streams start clogging the network, everybody's performance goes down. The only way for ISPs to deliver on their contractual bandwidth guarantees is to add more capacity—which kills their business model.

Calling All Bandwidth Hogs

Now a startup called Arootz, based in Netanya, Israeli, says it has devised a solution that greatly increases the Net's ability to deliver video without requiring substantial new infrastructure. The approach takes advantage of two existing (and not very esoteric) technologies: a standard Internet feature known as multicasting, by which the same information can be sent simultaneously to many recipients using just one data stream; and off-the-shelf hard drives that can store vast amounts of data fast and cheap.

The combination of the two represents a radical new approach to using the Net. Here's how it works: In today's world, when a person sitting at a PC clicks a link to watch a video, a series of servers delivers the content from its source to the end user in what's essentially an individualized person-to-person channel. If 10,000 people in a city all want to watch Desperate Housewives, each one of them gets their own personal bandwidth-hogging data stream.

Imagine, instead, that by keeping track of basic patterns and preferences, an ISP could anticipate that those 10,000 viewers might want to watch Housewives in a predictable time window—say, within 24 hours after each episode was released. So rather than delivering 10,000 individual streams whenever the viewers choose to watch the show, the ISP can blast it simultaneously via multicasting (using far less bandwidth) to capacious hard drives in all 10,000 homes, and then the customers can watch the show at their leisure directly from an in-home media server. "With storage becoming so cheap, we are downloading video content to users' personal hard discs so that it is immediately available regardless of network conditions," says Noam Bardin, chief executive officer of Arootz.

Getting Rid of Usage Bottlenecks

It's a bit like attaching a TiVo (TIVO) personal video recorder to the PC of every Internet user. After all, a 1-terabyte drive that now sells for about $350 is expected to hit $100 by next year and should get even cheaper in the future. Taking advantage of slow periods such as the middle of the night, ISPs could multicast all sorts of content to customers—highlights of yesterday's football matches, the morning news, or the top 100 most watched YouTube (GOOG) clips. All of it would be sitting there ready for access in the morning without putting any further strain on the network.

"The delivery is in off-peak hours but usage is on-demand, so we avoid the bottleneck that is occurring over global networks," Bardin says.

This approach also is a boon to ISPs.

"Video significantly changes their business model since it demands a continuous stream, which makes shared bandwidth difficult," says Danny Saar, manager of the Internet & technology sector at BBDO Consulting (OMC) in Tel Aviv. The high cost of adding additional capacity to meet demand has already led many providers to look for creative alternatives. Some are capping the amount of data that residential users get with their monthly plans, charging extra when the threshold is exceeded. Others are arbitrarily limiting speeds at peak hours.

Peer-to-Peer vs. Multicasting

The Arootz solution, developed by Yechiam Yemini, a co-founder of Israeli telecom software giant Comverse Technology (CMVT.PK), is based on a series of complex algorithms and management protocols that determine household demand and network constraints. The company figures that all of Israel could be served with just one server farm, while the U.S. could be covered with about 20.

By comparison, the king of today's so-called content distribution networks, Cambridge, (Mass.)-based Akamai Technologies (AKAM), has tens of thousands of servers scattered around the globe to distribute Internet content. Its biggest competitor, Limelight Networks (LLNW), has deployed a network of privately owned fiber to do the same.

Some companies, including BitTorrent and Kontiki (which was acquired last year by Verisign (VRSN)), are taking yet another approach. They're using a peer-to-peer (P2P) architecture, which spreads the workload by replicating content onto the PCs of thousands or even millions of PCs scattered around the Internet. The problem, Arootz contends, is that P2P doesn't address the fundamental congestion of the network, whereas using multicasting to minimize duplicate transmission does. Plus content still must be downloaded from peer nodes to end users when they want it—thus clogging the Net at peak times.

Shopping for More Capital

Arootz started initial testing of its technology in February in a trial involving delivery of hours of high-quality video content to 10 households in Israel. "We've seen nothing else out there like it that gives the consumer such high-quality television without utilizing the Net's full bandwidth," says the chief technology officer of a leading Israeli ISP who asked not to be identified.

A much larger trial of several hundred homes, in cooperation with some of Israel's leading content and service providers, is scheduled to take place this summer, with further tests in the U.S. and Europe set for later this year. To meet demand from Israel and abroad for the technology, the company aims to more than double its workforce, to 30, in the coming months. It's also looking for additional financing beyond the $5 million it has already raised from Israeli venture fund Gemini Israel Funds.

If all goes well, Arootz hopes to begin marketing its solution in 2008. The company is looking at two business models: either providing the technology to ISPs via licensing agreements and perhaps managing Net video services for them, or some sort of revenue-sharing model that would split ad royalties with content and service providers. Either way it looks like Arootz is on to something hot—and with little direct competition. Providers of Net-based TV services, such as Joost (which uses a P2P architecture) and Babelgum, are bound to be interested in what Arootz is up to (see BusinessWeek.com, 06/11/07, "Babelgum's Big Bet on Internet TV").

Now the company just has to prove itself. If it does, someday soon you could be enjoying smooth, broadcast-quality video delivered via the Net—without paying much higher prices for your connection or annoying everybody else in the neighborhood with your big fat downloads. That could make traditional broadcasters sit up and take notice.

Before it's here, it's on the Bloomberg Terminal.