Video-on-demand's great escape
Existing and looming VOD specifications and standards aim to pick the lock on proprietary systems, shave down costs, speed up deployments and, eventually, forge a common template for wide-scale everything-on-demand services
Buoyed by industry-wide standards and specifications already agreed to or currently underway, cable operators appear to be on the verge of breaking free of their proprietary video-on-demand prisons.
Today, MSOs that deploy VOD in a particular system are tied to one vendor, locked in by proprietary pieces of hardware and software. Early on, that wasn't a big deal, because it wasn't too taxing to integrate one set-top, one VOD vendor, one billing vendor and one content provider all on one cable system.
As cable operators extended on-demand services into other systems and began to explore more widespread rollouts, the integration of those components became more diverse and complex and, therefore, more time-consuming–and much more expensive.
With the billing piece alone, MSOs have the choice of employing one of several third-party billing systems or perhaps their own home-grown software. On-demand content providers, meanwhile, have their own unique treatment of metadata, which contains specific information about a VOD event or movie. Piled on top of that, VOD must integrate with each type of digital set-top and its program guide software, whether it's Scientific-Atlanta's SARA, TV Guide Interactive or Pioneer's Passport.
"Now we're looking at how we get to that next level of really going wide-scale for this as a service across many different MSOs and all of these different platforms," says Yvette Gordon-Kanouff, vice president of interactive technologies for SeaChange International.
That's the big challenge, and that's why common specifications and standards for VOD have become such a hot topic. CableLabs, for example, recently issued version 1.0 of its VOD content specification.
But that's just one piece of the puzzle. In order to truly open up cable's VOD network and to achieve the economies of scale necessary to roll the service out en masse, operators will require a bevy of common interfaces that aren't found in that initial specification.
Though it's not part of an official CableLabs initiative yet, Time Warner Cable's Interactive Services Architecture (ISA) back office specification, which is designed for VOD and other new services, appears to be one of the most likely candidates to help MSOs get to the point where they can more easily launch new services, without upsetting the applecart.
|Here comes Sun: Joining the VOD competition is Sun Microsystems and its Sun Media Appliance Platform. Pictured is Sun’s Netra T1 server for broadcast and media-on-demand configurations.|
"It may be OK today, but eventually you want to have a system where you could plug-and-play any server vendors," says Mike Hayashi, senior vice president of subscriber technology and advanced engineering for Time Warner Cable. That means the proprietary elements must be extracted like a dead tooth.
For an emerging service like "Maestro," AOL Time Warner's everything-on-demand project, peak usage is expected to surge to unprecedented levels. "For that to work...the cost of the stream must be equally small," Hayashi adds. "For the cost [per] stream to be very small, you have to have an environment where you could interchange pumps."
That day is coming. For now, though, industry-wide VOD specifications are tied strictly to content.The dawn of VOD specs
|Ride the bus: Broadbus Technologies hopes to challenge incumbent VOD players with a line-up of high-density video servers. The company’s B-1 server family is comprised of models for central- and edge-network storage.|
Originally, on-demand content was "bicycled" via digital tapes, and the metadata was imported manually. Managing metadata under a common thread will become critical as more content becomes available on-demand. While a movies-on-demand service might move 130 assets per month, SVOD could push that figure as high as 400, estimates Bob Chism, Concurrent's CTO and vice president of research and development. When personalized television applications or network-PVR services are added in, the metadata will flow even faster and in greater volume.
CableLabs is already fast at work on a follow-on, 2.0 version of the content spec, says CableLabs Chief Software Architect Ralph Brown. The next iteration will likely tack on capabilities for partial metadata updates for misspellings and other errors, instead of requiring the retransmission of the entire metadata package.
The 1.0 spec's encoding piece defines parameters such as video frame rates and bit rates. Encoding specifications make the process more efficient and cost-effective, says John Vartanian, senior vice president of technology and operations for InDemand, which taps third-party encoders such as CVC and VidFilm for its assets.
"This way, you encode content one time, and it works with all file servers and set-tops, and it spreads the cost across a wide base of cable systems," Vartanian says. Those costs can add up, he notes, as typical encoding costs for a two-hour movie can run more than $1,000. Still, the technical aspects of CableLabs' encoding spec haven't come without some misinterpretation and misunderstanding, notes Dom Stasi, chief technology officer for TVN Entertainment Corp., which encodes, aggregates and distributes on-demand content to MSOs.
According to the encoding spec, the threshold for absolute bit rate is 3.75 Mbps, and the limit on the video piece, sans overhead and audio, can be no greater than 3.18 Mbps.
TVN encodes material, scene-by-scene, at 3.0 Mbps. While that's below the spec's limits, it doesn't mean TVN is out of compliance, Stasi says. "If we go above 3 Mbps, we don't see a substantive improvement in [the visual experience]," he adds. "If we go below 3 Mbps on some [action-filled] scenes, we do see some just-noticeable degradation. So 3 Mbps, given today's state-of-the-art, is a nominal level for us."
When audio and other overhead data are included, TVN's encoding rate rises to 3.37 Mbps–still well within the spec's absolute bit rate limit. "A lot of people think it's not compliant, when it's absolutely compliant," Stasi says.
Encoding below the absolute threshold has its own set of benefits. "This accommodates more efficient use of server memory, more efficient use of bandwidth and more efficient ratios on the cable systems," Stasi says.Beyond content
To evolve today's proprietary VOD environment into a world of plug-and-play devices, several more standards and specifications must be added to the mix. Brown says CableLabs hasn't clamped down on any additional VOD specifications, although ISA has been the subject of some initial discussions.
ISA, while considered a de facto standard in some industry circles, is somewhat of a different animal for CableLabs, because it delves into business operations. Besides, "swallowing the metadata piece alone is a big enough challenge for now," Brown says.
Though ISA is cinched to the Pegasus platform, it also can be tweaked for Motorola Broadband systems. "The focus has been initially with S-A, but there's a lot of overlap on the Motorola side as well," says Raj Amin, senior director of business development at N2 Broadband, a key contributor to ISA's business management system interfaces. "We expect that there will be significant progress on that side in the next year or so."
Some of that is already done. Last month, Motorola and N2 Broadband unveiled a satellite-based, on-demand digital asset management and delivery system that melds N2's MediaPath system with Motorola's DigiCipher II uplink system and DSR 4500X Integrated Receiver-Decoder platform.
Another cable VOD specification effort could be linked to how set-tops communicate with the headend. So far, two protocols are apparently considered the most valid options: Digital Storage Media-Command and Control (DSMCC), which handles "trick-play" functions such as start, stop and rewind, and Real-Time Streaming Protocol (RTSP), a common standard. Streaming Service Protocol (SSP), another option already glued to ISA, is a pared down version of DSMCC.
CableLabs could, of course, take the liberty of selecting more than one approach, such as it did with DOCSIS 2.0 with A-TDMA and S-CDMA.
Another potential consideration is the 2.0 version of Pegasus' Asset Distribution Interface (ADI), which describes how metadata and the content are transported from a content provider to the MSO's asset management system.
Adding ADI 2.0 to the existing CableLabs spec "is just a natural," SeaChange's Gordon-Kanouff says. "The two go hand-in-hand."VOD vendors: The more the merrier?
Several "name" players have come to the plate, hoping to hit a home run with VOD as proprietary elements are pitched out. Silicon Graphics Inc., which participated in Time Warner Cable's ground-breaking Full Service Network trial in Orlando, boldly announced last month that it was "taking its gloves off" to grapple with incumbent vendors with its servers and storage systems.
SGI and its advanced imaging pedigree have yet to pierce the North American VOD market, although it has cobbled together some small deployments in The Netherlands, Spain and Taiwan.
Also joining the fray is Broadbus Technologies, which is offering high-density servers for cable and other types of broadband networks. The company has yet to score any U.S. cable deals, but has secured a non-exclusive and non-binding working relationship with StarzEncore.
Hewlett-Packard is another company aiming to join cable VOD's party with its hardware and software products tied to digital content management, content distribution and access devices. "We provide solutions to the marketplace; it's not just a box proposition," says Gabriele DiPiazza, worldwide broadband segment manager for H-P's network and service unit.
The most aggressive player attempting to wedge its way in is Sun Microsystems, which is trumpeting the Sun Media Appliance Platform, a suite of VOD software, servers, bit pumps and other hardware based on open standards and interfaces. Though Sun had yet to announce any VOD trials or deployments with MSOs as of press time, the company is optimistic about its chances.
"Overall, the response [to Sun] has been quite enthusiastic," says Rob Glidden, market development manager for Sun's broadband and digital media division. "We have not talked to a single operator that wants a proprietary lock-in solution. They're universally asking for a network open system."
Glidden contends that legacy proprietary VOD systems represent an easily surmountable hurdle, because, they're composed of "tweaks, twists and extensions" to existing open standards-based platforms. "These are little luggage locks, not padlocks," Glidden argues. "Yes, they are proprietary. Yes, they can be used to block a competitor. But these aren't major barriers to entry."
Incumbent VOD vendors, meanwhile, aren't exactly running for cover under the incoming hail of new competition. That's because, they argue, it's difficult to build a scalable VOD system without the proper service optimization. Besides, time is still on the side of the incumbents, says Concurrent's Chism. "In many respects, the war is going to be over 18 to 24 months from now. The runway is getting shorter."
Glidden counters that Sun has plenty of time to make a run at the cable VOD market. "You haven't seen mass deployment by the vendors," he says. "It's still a race to market, and I'd say we're in the horse race."
However, recent entrants such as Sun, IBM or SGI might have to acquire a legacy cable VOD company in order to quickly grab a solid position among the leaders, some industry observers say.
At least one appears to be willing and able. Diva, which laid off 20 percent of its workforce last month, has been on the block for months, sources say. At press time, a Diva spokeswoman declined to comment on whether the company is pursuing a buyer. Still, it's questionable whether a Sun or an IBM would be patient enough to stick around for what's considered a long-term pay-off.
"This is not a product or a service that's going to be a near-term, viable business for companies like Sun," says Sean Badding, vice president and senior analyst for The Carmel Group. "But they do realize there's a lot of potential in VOD. [Incumbent VOD vendors] are talking about successes, but this will take more time than people are expecting."
"The movie-on-demand server market is pretty small," adds Time Warner's Hayashi. "But if you add more on-demand content, which then means more streams, more pumps, then the market grows. Then I think [new VOD competitors] will be making business decisions to show up."Plug-and-play VOD
Showing up and making it worth the effort hinges on whether their equipment and software can fuse with legacy systems.
Though there's some disagreement on this subject, many in the industry see VOD specifications and standards as a path paved for off-the-shelf, plug-and-play components, including the video servers. However, several elements need to plug into the backend in order for the cable network to become truly open. They include the billing mechanism and network resources such as video server stream capacity.
"Those are all pieces that have been done in a proprietary fashion in the past," says N2 Broadband's Amin. But, with a standard interface, everything–from the content, video servers, gateways, applications and billing systems–would plug into the business management system. That, at least in theory, would allow a cable operator to pick and choose from a large menu without suffering the integration headaches.
N2 Broadband tested that concept about a year ago with its MediaPath business management system. The company switched and swapped ISA-compliant servers from Concurrent, nCUBE and SeaChange to play movies on digital set-tops in a lab setting.
Those tests passed "with flying colors," Amin says, but adds that the process operates well in environments with master headends, but might not as easily bust through the proprietary tie-ins associated with distributed VOD architectures.
Although the VOD server plug-and-play model is a promising theory, "realistically today, it's not going to happen, because of the [proprietary] propagation piece," Amin says. "But that's being handled in the next revision of ISA."
However, SeaChange's Gordon-Kanouff disagrees with the notion that the plug-and-play scenario also refers to video servers, which are tailored specifically for integration with on-demand services on a broadband network. Adding specifications and standards "doesn't make our equipment plug-and-play," she argues. "It just makes it easier for us to work with all of the different non-VOD components that MSOs work with."
Whether VOD components can become plug-and-play is subject to interpretation, offers Naresh Makhijani, Diva's general manager of on-demand systems. "If you define the server as the box, then I think the answer is no," he says. "If you define the server as the server system, and you have interfaces for billing and content and applications that are commonly defined, then, as a system, you should be able to replace one for the other."
If the playing field is leveled out in such a way, incumbent VOD vendors remain confident that they can differentiate themselves beyond price via powerful backoffice and business management system software and scalable, but efficient, architectures. "Most of our energy and a lot of intellectual property is poured into the software," Makhijani says. "That means we're able to take pretty much any brand name server–Compaq, Dell, IBM, H-P–and install an open Linux [operating system] on there and put our software there and optimize it as a very efficient streaming engine."
"VOD is not about servers," says nCUBE Corp. President and CEO Mike Pohl. "It's about the ability to wrap around all of the business logic and everything else you need to run that business."
"The real value is on the software side," Chism agrees, noting that vendors attempting to muscle their way in could be "multiple years behind in that area."
"You just can't talk about getting cheaper," says Pohl, "because you're looking at scale, and how to store thousands of hours versus hundreds of hours and how to manage those thousands of hours."