Google Cloud Goes After Media & Entertainment Customers
Google is going up against Amazon, Microsoft and IBM to make its cloud services the platform media companies should look to for their needs. It has built the largest private network in the world, which is now available for external customers. "Google thinks media is a very important vertical and deserves special attention," said Leonidas Kontothanassis, engineering director for Google at his Content Delivery Summit keynote in New York on Monday.
Approximately half of Google's engineers are working on the various components of Google Cloud Platform—the largest engineering group at Google by a factor of three, said Kontothanassis. Google Services provides an entire platform, powering media customers like Snapchat, which runs on Google Cloud.
Google has invested $29 billion for cloud infrastructure in the last three years. "We are serving a billion unique IP's every day. We bring the highest reliability in the industry," says Kontothanassis. Google Services for the platform include APIs for machine learning, specific custom services providing up to 5,000 cache points, SSL delivery at no additional cost, and a range of other services.
Services they are providing for media include rendering, processing, intelligence, monetization, and the playback platform. The promise: Google Cloud can provide faster time to market and greater costs savings for media production, plus help customers avoid congestion and security problems of the public internet.
Kontothanassis outlined a couple of Google use cases. A live two-hour sports event can now be done for $4,000 for virtual live linear delivery with no hardware costs. Spotify can now process data in 15 minutes that previously took 96 hours, enabling them to provide customized personalize live linear music streaming. "Personalization by far is the most interesting development and most ripe for disruption," he said.
Video Supply Chain
Machine learning is another area Google is providing access to with API's. Live streaming can take advantage of real time closed caption creation in 80 languages. Image recognition can automatically identify logos and objects within video content which can be paired with customized ad insertion based on content or other monetization options like dynamic links to ecommerce sites. Intelligent playback can allow viewers to create customizable user defined clips (where a viewer can request highlights for a specific athlete and specific game for example). AI can be used to do analysis of video compression to allocate optimum file sizes based on traffic and delivery data. Any combination of these services provide huge opportunities for media innovation.
Google provides for demand, capacity and congestion changes in real time. "Our workload is 20–25% of all internet traffic, depending on the country," says Kontothanassis, and 90% of this traffic is YouTube. "One of the most difficult things to do is how do you decide which users get assigned to which section." Every client gets mapped individually. Google's content mapping system provide individual optimized delivery, and can remap the client if they decide they need to move traffic.
Cedexis has generated independent reporting on Google latency, availability, and throughput.