Editor’s note: Dror Gill is CTO of Beamr, an imaging technology company that powers web publishers, digital distributors, social networks and media companies and focuses on reducing costs associated with storing and transmitting media files.
Today’s consumer is streaming more video than ever before. Netflix alone has increased its subscribers by more than 20 million in the past two years, totaling 50 million subscribers as of Q2 2014. It’s no wonder, then, that the increase in video consumption is not without consequences for streaming services and ISPs.
The news media has recently highlighted a drawn-out dispute between Verizon and Netflix, with Netflix blaming Verizon for its congested network and failing to provide the consistent bandwidth required to enable a high-quality service for Netflix customers. Netflix even sent messages to its customers blaming their poor experience on Verizon’s networks.
Eventually, Verizon filed a cease and desist that forced Netflix to withdraw these messages, with Verizon blaming Netflix for continued traffic problems due to its limited choice of transit providers.Verizon’s VP of regulatory affairs, David Young, even showed the following diagram in a blog post, claiming there is no congestion on Verizon’s own network, only on the link from Netflix to Verizon through their choice of transit providers.
The debate has continued with Netflix claiming that Verizon should upgrade its connectivity to these transit providers to solve the bottleneck issue. Young immediately replied that Netflix should use paid peering and pay Verizon for the inbound traffic flowing into its network. And the battle goes on.
This dispute is just the beginning of the Internet bottleneck issue. Not only do on-demand video services such as YouTube and Netflix now account for over 50 percent of all North American Internet traffic, but Cisco recently reported in its Visual Networking Index report that by 2018, video will comprise a whopping 79 percent of global consumer Internet traffic.
Whether or not streaming services and ISPs eventually sign peering agreements to settle the disputes, there are still several possible solutions to the video bandwidth problem. An obvious one is to reduce video bitrates in order to lower the bandwidth requirements of the streamed video files. However, because video quality is directly related to the bitrate allocated to the video stream, blindly lowering the bitrate will result in a poor viewing experience and unsatisfied customers—an option that is unacceptable in the age of retina displays and UHD 4k televisions.
Another solution is caching the most frequently viewed video files at the network edges. This ensures that when a popular video file is being requested by a user, it can be streamed from a location that is close to the user’s physical location, and does not have to travel again over the Internet backbone. Since most of the online video traffic is generated by a relatively small number of popular streams, caching those streams can be cost-effective when taking into account the storage costs of the cached files vs. the delivery costs of each copy that travels over the network.
Adaptive bitrate streaming is another common solution used by content delivery networks. This method detects a user’s bandwidth and CPU capacity in real time, then adjusts the quality of a video stream accordingly to prevent overloading a user’s connection with more bitrate than it can handle. While this strategy incurs additional storage and encoding costs, it can eliminate buffering and provide consistent streaming on both high-end and low-end connections.
Finally, there’s media optimization, which takes an already-compressed video stream, analyzes its perceptual properties, and encodes it to a lower bitrate to increase streaming speeds without affecting the original video quality. This would be like taking a ball of modeling clay and squeezing it to make it smaller: It still has the same amount of clay, but occupies a smaller amount of space. Some forms of media optimization may struggle to maintain the quality of the video while reducing file size, but when done correctly using a reliable perceptual quality measure, this process can reduce a bitrate and file size by 20-50 percent while retaining the full perceptual quality.
While these major players continue to sort through the congestion issues, utilizing current solutions like caching, adaptive bitrate streaming and media optimization can alleviate the bandwidth bottleneck problem while providing a win-win-win situation for content providers, telcos and end users.