Monday, April 18, 2011

Mobile Video & the Significance of Adaptive Bit Rate - A Simple Explanation

[This is a guest post by David Dudas, Sorenson Media's Vice President of Video Solutions, that first appeared in BusinessInsider a few days back. This is David's second thought piece in BusinessInsider in two months -- his earlier piece explained the significance of the H.264 v WebM format debate. David takes a decidedly non-techy approach to these discussions, which makes them very useful indeed to the business executive, no matter how techy they may think they are. In other words, his thought pieces give real-world practicality and tangible understanding.]

It's not discussed often, but online video is undergoing a quiet transformation that's fundamentally altering the economics of digital distribution. What is this radical transformation? It's not over-the-top delivery, a new video-on-demand (VOD) service, or even a new tablet or smart phone. Instead, it's something much more mundane: the adoption of Adaptive Bit Rate technology.

Adaptive Bit Rate, or ABR, is a technology that allows online video to be viewed without frustrating interruptions for "buffering" – something we've all unwillingly experienced. Waiting. Waiting. Waiting. Simply put, ABR enables uninterrupted playback by constantly analyzing the viewer's Internet connection and "adapting" to poor bandwidth conditions by switching to a lower quality stream. It's the digital equivalent of driving home from work on city surface streets when freeway traffic is at a standstill.

As recently as two years ago, ABR was considered a nice-to-have feature for content creators and online video service providers. The vast majority of online video was consumed in-the-home or at-the-office on PCs with decent broadband connections. While the average U.S. broadband speed of 3 Mbps paled in comparison to South Korea's 15 Mbps (and still does!), it was good enough to view online video that was delivered via now-outmoded technologies such as progressive download and constant bit rate streaming.

Today, everything is different. It's no longer guaranteed that an online video viewer is attached to a wired broadband connection. Today's viewing audience is just as likely to be streaming video to an iPhone or Android device over a 3G wireless network, or to a set top box in a home where three other people are saturating the network by streaming video to their iPads and laptops. In other words, the diversity of today's streaming environments and proliferation of viewing devices makes ABR a must-have technology for content creators and service providers, not a nice-to-have. The trends overwhelmingly point to greater device proliferation in the future. Remember, the tablet category of personal computers didn’t even exist as recently as one year ago.

So the user experience is all well and good, but how does Adaptive Bit Rate alter the economics of digital distribution? To answer this question, it's helpful to understand that ABR avoids buffering by delivering very short video segments (typically 5-10 seconds) – one after another after another – to the viewing device rather than one continuous stream or one large file. Every five seconds the viewing device requests the next video segment from the media server. In response the media server delivers a lower quality (i.e. lower bit rate) segment if the connection speed is degraded, or a higher quality segment if the connection is sufficiently fast.

To enable this adaptive form of delivery, content creators and service providers must chop their videos into a bunch of five-second segments. Then, they need to create multiple "renditions" of every five-second segment – e.g. a 200 kbps segment for poor bandwidth, a 400 kbps segment for slightly better bandwidth, and so on. To illustrate, let's assume we want to prepare a two-hour movie for 10 different bandwidth conditions. First, we chop the video into five-second segments:

120 minutes ÷ 5 seconds = 1440 segments

Then, we create 10 renditions of each 5-second segment:

1440 segments X 10 renditions = 14,400 little digital files

In other words: you will need to transcode and store over fourteen thousand digital files for every two hours of source content.

Unfortunately, the madness doesn't stop there. The 14,400 files described above may look great on an iPad, but will be completely incompatible with Android devices or devices that use Flash or Silverlight for playback. So one must transcode and segment several additional renditions to support the growing number of smart phones, tablets, and connected TVs flooding the market this year. Multiply all those renditions by the number of titles in your catalog....Wow, that's a lot of transcoding and storage!

In a nutshell, THAT is how Adaptive Bit Rate alters the economics of digital distribution. In years past, transcoding and storage were manageable expenses because a single source video was typically transcoded to a single output file. In the age of ABR, however, the sheer volume of output files requires significantly more compute power and storage space. Clearly, content creators and service providers cannot rely on the traditional hardware model to support ABR unless they want to blow their IT budgets on truckloads of servers and teams of engineers to keep them running.

Fortunately, there's a place where high-volume, high-complexity ABR transcoding makes economic sense: in the Cloud. As content creators and service providers migrate their transcoding workflows to the Cloud, they quickly discover a myriad of advantages over the traditional hardware model: drastically reduced CapEx and OpEx, on-demand elastic capacity, automated scalability, increased agility, quicker time-to-market, and built-in future-proofing. In fact, the economics of video transcoding in the Cloud are so advantageous, content creators and service providers who try to build or maintain in-house systems will find themselves at a debilitating disadvantage to their competitors.

If not for Cloud computing, the cost and complexity of Adaptive Bit Rate would kill online video. Providers would either stick with outmoded technologies – which would discourage consumption due to the maddening buffering experience – or providers would drain their resources on unsustainable, traditional hardware infrastructure. Luckily for all of us digital media junkies, the Cloud is here to rescue online video. This is a case in which you really can believe the hype.