The obstacles to scaling multi-screen TV are not insurmountable. Netflix clearly has figured out how to deliver movies and previously broadcast TV content to an extensive range of devices. The number of worldwide service providers that have launched on-demand TV Everywhere-type services has grown by a factor of ten (from five to more than fifty) over the past three years, according to Parks Associates.
Some providers also have plunged ahead with live multi-screen TV offerings, such as New York-based Cablevision, which launched an iPad app last year supporting nearly 300 linear channels. This week Time Warner Cable announced a live TV to PC (and Mac) application, having already deployed live linear for the iPad and iPhone. Hong Kong-based PCCW has been at it for years, with its IP-based Now TV.
But obstacles remain. There are in-home limits in U.S. markets and other constraints subject to deals with content owners. “It is a very content-sensitive landscape,” said Comcast SVP Product Engineer Mark Muehl. But the challenge is not just a matter of negotiation.
While dealmakers sort through contracts, engineers consider the implications of going even further. To replicate the entire lineup that Comcast broadcasts into its markets today, for instance, would involve about 10,000 unique linear channels, including all the public, education and government (PEG) channels. “That’s obviously a lot of rights relationships to manage,” Muehl said.
Beyond operations, there are capacity questions. Muehl said that one way to assess the last-mile issue is to look at data for the percentage of Internet traffic consumed by Netflix and the time its subscribers spend using the service, and then compare that to the amount of video that a cable subscriber views each month.
According to data released last October by Sandvine, Netflix was peaking at more than 32 percent of Internet traffic. Estimates for monthly hours Netflix subscribers spend vary, but it is clearly some fraction of what cable video subscribers view. “When we think of the prospect of winding down cable TV service into the home and replacing it with a unicast stream of unique content for them over the home, the numbers get staggering, very quickly,” Muehl said.
Are there ways to reduce the load of Internet video? Yes, which is why encoding start-up EyeIO, which came out of stealth mode in early February with the announcement that Netflix has licensed its technology, is attracting attention. As reported by Fierce Online Video, the company was in New York on the day of the announcement, visiting with other prospects. EyeIO Chairman Charles Steinberg (formerly of Sony) told editor Jim O’Neil that a list of potential customers would be long, including “anyone who want(s) to delivery quality video and dramatically use less bandwidth on it.”
How the technology works remains a bit of a mystery. CEO and CTO Rodolfo Vargas described his software as an encoding 2.0 engine that uses “artificial intelligence” to sense or view the video coming in and going out. (Hence EyeIO.) Unlike encoding techniques that only “massage the video” from the top level, this engine deals with the “elementary streams.” As such, Vargas said it stands apart from MPEG segmenting implementations, such as HTTP Live Streaming (HLS) or Dynamic Adaptive Streaming over HTTP (DASH).
Vargas nonetheless underscored the importance of MPEG-DASH. “Everyone has been waiting for this standard,” he said. The promise of the M-DASH, which was ratified in December but continues to be shaped by the MPEG community, is to simplify HTTP video transport to two options: segments using multiplexed MPEG-2 transport stream (TS) or elementary streams using fragmented (f)MP4 files.
Netflix is an M-DASH supporter. According to a Netflix spokesman, the company has been using fMP4 files since the beginning of its streaming service in 2007 and has been engaged in the M-DASH standards efforts since September 2010. “Our media files are already DASH-compliant, and we see DASH as having immediate value as it standardizes many device capabilities that can used by Netflix stand-alone streaming applications,” the spokesman wrote via email.
“They definitely have a noble goal,” Muehl said, of the M-DASH initiative. “Our concern is that we want to make sure that any solution to standardize along those lines really does address the core problem.” If the standard is just a wrapper around proprietary implementations then Muehl is “not sure that this really hits the nail on the head.”
“If we were to be asked, we’d want HTML-5 to be a programmable or scriptable pipeline,” he said. “We’d like clients to be smart, in a programmable sense, to ask for the right kinds of things based upon circumstances.”
For more on Scaling Multi-Screen TV, please see the forthcoming Videonet report on the topic.