Discovery Accelerates Innovation, Cuts Linear Playout Infrastructure Costs by 61% on AWS

Global media company Discovery needed to innovate its infrastructure for linear playout programming to maintain its leadership position in a rapidly-changing industry. Although it was reaching viewers through new channels and streaming services, Discovery used on-premises software for its playout infrastructure, which transmits content to consumers through broadcast, cable, and satellite. The cost and time required for the upkeep of its on-premises playout was significant, slowing growth and innovation. 

In 2016, Discovery began to migrate its workload to Amazon Web Services (AWS). The company modernized its playout infrastructure using Amazon Elastic Compute Cloud (Amazon EC2), which provides secure, resizable compute capacity in the cloud. By removing the need to set up full-scale data centers, Discovery was able to decrease costs and improve reliability for affiliates and content consumers. Now, Discovery can scale to deploy new channels faster, grow and innovate, and pivot to meet the needs of an evolving market.

Every year, Discovery delivers over 8,000 hours of original programming in nearly 50 languages to 220 countries and territories. Its offerings include HGTV, Food Network, Animal Planet, and Discovery Channel. The company’s portfolio of networks reaches viewers through cable, direct-to-consumer streaming, digital channels and providers, and more. 

Discovery previously managed several full-scale data centers for its global playout system. However, the data centers were expensive because Discovery had to make a significant upfront capital investment in servers and incurred sunk costs when full capacity wasn’t needed. Upgrades were manual and so time consuming that Discovery could only perform up to two per year. The company ran redundant servers to improve availability, but because they were housed in the same facility, it reduced overall fault tolerance. 

One reason Discovery migrated its global linear playout system infrastructure to AWS was the need to open a major new data center in Europe. By building in the cloud, Discovery avoided paying for fixed infrastructure, reducing its on-premises server footprint from 130 racks to 10. “Investing in a major data center facility requires us to think in terms of an investment that will last 20–30 years,” says Dave Duvall, chief information officer of Discovery. “We can’t imagine feeling certain about the state of the linear playout market for that length of time.” Discovery completed the migration of its global channel estate to the cloud in June 2021.

Comprising the foundation of Discovery’s playout infrastructure are Amazon EC2 and Amazon Simple Storage Service (Amazon S3), an object storage service that offers industry-leading scalability, data availability, security, and performance. The playout system uses about 1,000 Amazon EC2 instances, plus instances for testing and quality assurance. Amazon S3 stores 15 PB of content at any given time. Discovery cut the total cost of ownership by 61 percent on AWS. That’s partly because of an Amazon S3 Storage Class called Amazon S3 Intelligent-Tiering, the only cloud storage class that delivers automatic cost savings by moving objects between four access tiers when access patterns change, as well as Amazon EC2 Reserved Instances, which provide up to 72 percent discount compared to Amazon EC2 On-Demand Instances. 

Discovery also uses AWS Direct Connect, a cloud service solution that makes it simple to establish a dedicated network connection from on-premises systems to AWS. It transmits playout streams from Amazon EC2 to the company’s on-premises distribution facilities in London and Sterling, Virginia, as well as partner uplink and teleport facilities globally. 

Through the automation and managed service of AWS, as well as the support of AWS Partner Evertz, 80 percent of Discovery’s engineers can focus on innovation and automation efforts delivered through software, compared to 50 percent in the past. Discovery’s software automation can monitor playout signals based on errors and exceptions, drastically reducing the need for constant human monitoring. Employees who once monitored four channels now monitor 40–60 channels, so Discovery can now roll out new channels without increasing staff headcount. Discovery cut the time to launch new services by 50 percent because being on AWS makes it simple to test and launch the configuration. 

Using a multiregional model of redundancy on AWS, Discovery gains disaster recovery and high availability and can mix workloads across regions as needed. A server running in a US AWS Region has a backup running in a European AWS Region. Previously, if hardware failed, the company would have to run without redundant capability until the vendor could provide a replacement, which could take weeks. Using automated deployment, Discovery instantly replaces failed instances. 

Discovery often performs multiple production upgrades in a single week on AWS. “We’ve upgraded all our global video playout chains over a multinight maintenance period using automated deployment, with no disruption to customers,” says Ron Yoslov, group vice president of broadcast engineering and operations at Discovery. As a result, customer video experience has improved, with interruption decreased by 97 percent, from 240 seconds of interruption per week to 7 seconds. “Through the service-level agreements provided by AWS and the connectivity plan we’ve invested in, we feel comfortable running our multibillion-dollar business on a two-site architecture,” says Duvall. 

Using automated deployment, the company switches instance types in a matter of hours or days. That agility was essential when Discovery had to shift licensing for a group of channels from the United Kingdom to Amsterdam after Brexit. The company was able to test the Amsterdam licensing configuration before deploying it, which would have required a costly and time-consuming change management project and the provisioning of redundant physical infrastructure if performed on premises. “We pay minor On-Demand Instance dual-run costs, but it’s a fraction of what it would cost to do it in on premises, and we have the confidence that it’ll work because we can test it as much as we’d like,” says Duvall.

Discovery can now deliver a more reliable service to customers while cutting costs and increasing agility and staff productivity, helping it remain competitive in the media industry. The company will continue to take advantage of AWS services to support the creation and delivery of content worldwide. Duvall says, “Our media processing needs the scalability of AWS to be effective, especially as direct-to-consumer markets are the driver of our growth.”