Jason Gaedtke is director of engineering at YouTube and is speaking at the Terabit 2013 conference, taking place at the Sheraton Fisherman’s Wharf Hotel on 4-5 June in San Francisco. Ahead of the show we speak to him about YouTube’s strategy for coping with current demand and what the latest developments are as it plans for future growth.
In terms of infrastructure what do you have to put in place to support hugely popular videos and events such as the Olympics and what are your plans for future growth?
Fortunately at Google there are a lot of ‘tall-shoulders’ to stand on when you are deploying at scale. Horizontal build-out on commodity hardware is fundamental. YouTube’s video infrastructure runs on the same cloud that powers other Google services, reaching every corner of the globe. Another strategy for the Olympics was developing a system than anticipated and even expected localised failures, with built-in redundancy and convergence times that were not noticeable by viewers. As for future growth, people are watching so much online video that it’s possible they may one day watch more videos than the current capacity of the Internet. As the old saying goes, watch this space.
Where do you see as the biggest growth areas for YouTube video views in the next few years?
People traditionally watched YouTube through a browser – mostly videos found from search and link-sharing. Recently, people are finding more content they love on YouTube by subscribing to channels, and they end up watching more video on every web-connected screen they have. This is driving significant growth in mobile consumption through native apps on smartphones and tablets. Based on current trends, we can see both of those platforms continue to grow, so we’re investing heavily to optimise those experiences, especially around subscriptions.
Stopping there would neglect the fact that we all spend large amounts of money and effort crafting ideal viewing environments within our homes, with large-screen displays, high-end audio, ambient lighting and, importantly—snacks! YouTube’s migration onto devices like game consoles and connected TVs will have significant impact on this overall ecosystem, including content production, consumption and commerce.
The Terabit 2013 conference is taking place on the 4-5 June in San Francisco. Click here to learn more about the show.
Despite recent lobbying in Europe, European Union budget funding for broadband networks was not forthcoming. Is the pace of the rollout out of fast connections worldwide a concern for you?
People are watching more and more video—at a terabit scale—and there’s a symbiotic relationship between the growth of broadband networks and YouTube’s ability to meet on-going demand. Access network technology, capacity and pricing vary widely around the world, so we have to tackle lots of challenges to give people a consistent, awesome experience on YouTube. In some areas we see healthy local competition, which keeps price/performance competitive and ultimately benefits both users and services. In other areas, access technology is physically constrained or perhaps offered by only a single provider. Those markets are typically more challenging as there are fewer options for pricing and performance.
From your perspective what is the most exciting internet technology development for 2013? Is it the HEVC codec or something else?
Advanced codecs are definitely compelling, bringing 40-50% compression efficiency at equivalent quality versus today’s standards. This means better performance, quality and cost-reduction for both consumers and service providers while transmitting fewer bits. I’m also excited about what’s happening on VP9 within the WebM community. This advanced codec looks to have similar or better performance characteristics to HEVC, while also being developed under an open, royalty-free model to spur more innovation.
One of the technical areas that I’m most excited about is high-performance, low-latency video delivery. We have a program underway at YouTube where multiple teams are swarming on this problem and already showing progress. This has the potential to fundamentally change the way we deliver and consume online video, going forward.
Do you think that mobile network operators are fully prepared for the data explosion that’s coming?
Local spectrum management, which involves technical, business and public policy considerations, and basic economics, can be significant limiting factors to widespread deployment and consumption of mobile bandwidth through video.
Micro-cell technology offers some relief from a technical standpoint (increasing density of wireless radios and localising spectrum usage), though consumption-based pricing models remain 100x those of wireline services on a per-GB basis. Many tablet users, consequently, end-up shifting video consumption to WiFi-enabled wireline services to avoid these fees.
I’m hopeful that we’ll start to see increases in monthly consumption tiers and declining consumer pricing, as the technology matures and cost-reductions become available. We are also interested in working directly with wireless operators to increase efficiency and quality of experience over their networks, while managing costs.
Finally, can you dance Gangham style? 😉
Sure. I’m from Colorado and we have a few horses around so lots of those moves are familiar. Also, Korean-rap is huge there. One thing I can’t do is bust-out the Harlem Shake. We’ll need some strong audience participation at TeraBit to help out there… Who’s got the banana suit?