HD and VOD: Ever the twain shall meet
Peanut butter and chocolate. The Captain and Tennille. Fred and Barney. These are but a few pairs that go together as naturally as high- definition television and video-on-demand.
A handful of MSOs already have HDVOD plans in the works. Cablevision will be the first MSO out of the blocks late this summer when it launches a commercial HDVOD service tier. Time Warner Cable, already the industry leader in terms of HD and VOD market deployments as well as HD-capable set-tops shipped (120,000 at last count), is positioning itself for HDVOD, as well.
"We're actively testing it now," says Mike LaJoie, Time Warner Cable's executive VP, advanced technology. LaJoie wouldn't say when and where the MSO will put HDVOD in front of customers, but noted that Time Warner has gone as far as creating a deployment schedule.
But before HDVOD services can be rolled out en masse, it's clear that operators must upgrade or enhance their existing, standard definition (SD) VOD systems or set aside traditional thinking when architecting their new ones.
For starters, those systems must accommodate much larger video assets– somewhere in the range of four to five times that of SD. Those massive files can render chokepoints and spawn trouble up and down the VOD system–from the content's origination, down to the catcher, and on the transport network into the consumer's home.
Because those points of contention exist and still require some ironing out, operators are taking a deliberate approach to HDVOD, starting trials with enough content to make the service interesting, but not heavy enough to bring the system to its knees.
"If you went out there with 50 HD movies, you could cause some trouble right out of the chute," says Mark Hess, VP of digital television for Comcast Cable. "So we plan to start slowly and manage the networks accordingly in order to provide a high-quality service."
But most of the primary pieces of HDVOD are in place. "Once you do SD in VOD, you've climbed 95 percent of the hill to offering HD," says Joe Ambeault, director of broadband systems for VOD vendor SeaChange International.
But completing the final five percent of that journey still requires some doing.The Art of the Encode
The magic of the encoding process is to pack enough of the "right" information into the space given. Adding the right information ensures that the content maintains a high-quality look and feel, devoid of artifacts that the average eyeball can detect.
At 3.75 Mbps, CableLabs and the cable industry has already specified the sweet spot for the encoding of standard-def VOD assets. It won't be quite as straightforward with HD.
Encoding HD content presents a problem not as evident in SD. An HD drama, for example, might require fewer bits per second than would a football game or a high-action movie and still look great.
"We haven't determined the sweet spot yet," says John Vartanian, senior vice president of technology and operations for InDemand, which aggregates and distributes VOD content for several major U.S. MSOs. He says early experiments suggest that the numbers vary, with 11 Mbps appropriate on the low end, and 16 to 17 Mbps for movies and other top-tier programming.
CableLabs is presently working on a new encoding spec for HDVOD, but nothing's been determined yet. Vartanian notes that a working spec should emerge in time for MSOs to take HDVOD for a test drive by the fourth quarter.Making the Pitch
Once HDVOD content is encoded, it must then be pitched to the cable headend. If not engineered properly, it can become an expensive headache if assets with errors must be re-pitched in their entirety.
InDemand and N2 Broadband, a maker of VOD pitchers and catchers, recently concluded the first phase in a series of HDVOD lab tests. The first part of the trial tested software modification that allowed the N2 catcher to accept files larger than 8 GB, a figure that marked the limit for SD catchers. InDemand then encoded an HD movie, loaded it onto the pitcher, and transmitted it via satellite to an N2 lab in Atlanta. Parallel to that, the companies also pitched HDVOD to a Time Warner Cable lab in Charlotte, N.C.
With that part of the test deemed successful, the companies plan to conduct the next phase of testing, which will demonstrate how to pitch an HD asset in chunks and reassemble them when they reach the VOD catcher. That's an extremely important capability because re-pitching an entire HD movie file is costly, should errors crop up in the metadata or in the content itself.
Given that an HD file will run on the order of 8 to 12 Mbps on the cable system, the same bandwidth issues need to be considered when an asset is pitched via satellite. Time is money.
In today's SDVOD world, a company like TVN Entertainment Corp. has about 750 real hours each month to distribute between 1,000 to 1,500 hours of content per month via its transponders. "If we're doing things in half real-time, every two-hour movie would take one hour to transmit," explains TVN CTO Dom Stasi. If TVN decided to double the bandwidth again, it could send out 3,000 hours of content in a 750-hour month.
Fortunately, distributing VOD is much more efficient than sending out pay-per-view/near-video-on-demand (NVOD) titles. "If you're watching a movie at home off an NVOD feed, we can't send it twice as fast because people won't be able to see it," Stasi says. "But with VOD, we can crank up the throughput and bandwidth."
But, with standard VOD libraries continuing to grow and HDVOD in the wings, even today's systems will eventually require an upgrade. "We believe we'll have to increase our capacity this year," Stasi says, noting that NVOD presently takes up about two-thirds of TVN's transponder capacity, with the balance going toward VOD. "If VOD traffic gets heavier, which I think it will, we're going to have to decrease our NVOD traffic or acquire another transponder."
On the other end is the VOD catcher, and that piece will also need some upgrades, lest it become a bottleneck. "We will have to expand catcher storage," says Raj Amin, VP of business development for N2 Broadband, which this year launched an enhanced version of its Mediapath platform to handle HD.
Standard N2 catchers come equipped with one 72 GB drive, but each catcher contains expansion slots to double or triple storage capacity. Future versions of the Mediapath platform will provide more standard storage, he says.Serving Up the HD
VOD architecture also will play a key role in determining the efficiency of offering HD content. Centralized storage will also become en vogue as operators deploy HDVOD. "You have to plan on five times the bandwidth," says Gary Southwell, VP of marketing with Internet Photonics.
By using a centralized architecture, "I'm saving storage costs as much as 95 percent," adds LaJoie.
Major providers of video servers believe their equipment can store SD and HD content on the same server simultaneously without too much trouble.
But an HD program can take up to five times more storage space than its SD counterpart, so much more storage might be required. But until HD sets reach critical mass, streaming requirements won't climb the ladder as fast as the storage. Vendors say they have addressed this issue by decoupling the streaming from the storage.
HDVOD "hits the storage side before it hits the streaming side," says Bruce Bradley, vice president of product management for Concurrent Computer Corp.
nCUBE Corp., another VOD server and system provider, says its new n4X servers and previous generation N4 systems can stream HD content today. "Now what we're doing is fine-tuning how to maintain rated throughputs with SD and HD from the same hub," says Terri Richardson, VP, product management and business development for nCUBE. A fully-loaded nCUBE server can store north of 170,000 hours of SD and between 40,000 to 50,000 hours of HD content.
Some younger startups believe they have something special to offer, as well. Broadbus Technologies, for example, believes its DRAM storage approach provides a more efficient approach to HDVOD than traditional disk-based, RAID set-ups because its product is made to handle massive HDVOD assets that might be in the range of 20 to 25 GB. Plus, that content doesn't require replication.
"We can share that [HDVOD] asset across multiple streams," says Broadbus CTO Tom Jokerst. The Broadbus system, he adds, doesn't need extra room to store trick modes (fast-forward, rewind, etc.). By supporting trick modes on the fly, the Broadbus platform conserves about 20 to 30 percent on overhead.
A fully-loaded Broadbus server can send out 19,200 SD streams at 3.75 Mbps, or, in an HD world, between 4,000 and 5,000 streams at about 20 Mbps.
The HD program also needs a wide pipe to get to the consumer's home. Many early VOD systems used DVB-ASI as the transport, but they'll be of "very limited" use for HDVOD, says Southwell. Gigabit Ethernet will be the ticket for HDVOD, he adds.
LaJoie believes, though, that it won't make much difference initially: "You can layer MPEG streams on top of ASI just as easy as anything else." But as the number of HD boxes and the amount of content grow, he acknowledges that it might warrant the need to optically segment VOD bandwidth.ROI and 'The Why'
Even with engineering tweaks and proper pieces in place, offering HDVOD will still cost a bit extra. For one, per-stream costs for HD will be higher than those for SDVOD, which have plummeted to the realm of $300.
So, who's going to pay for it? The consumer, most likely. It's a good bet that those HDVOD titles will carry a premium price. But jacking up the price is just one method to recoup those costs. Another is advertising, an element that should be used before consumers get too used to getting beautiful HD content for free.
The why is much easier to resolve. HDVOD clearly could give cable an upper hand against DBS and the video store. "We've got it [HDVOD] before Blockbuster; that's a big one," grants David Ludder, InDemand's senior director of technology. "We'll have it before the rental companies or before Netflix," he says, referring to a burgeoning firm that distributes DVDs through the mail.