Playtech Improves Reliability by Delivering Low-Latency Video Globally Using Amazon CloudFront and Lambda@Edge

Founded in 1999, Playtech supplies gaming software for online poker rooms, sports betting, and more. The company’s service depends on delivering low-latency, high-quality video streams to players across multiple video protocols and various geographical locations simultaneously. Previously, it used a combination of origin and delivery products across several different suppliers, but that solution was difficult to manage and did not scale efficiently. The company experienced streaming challenges that made it difficult to roll out new features and expand into new markets.

To promote growth and innovation and to improve the reliability, availability, and efficiency of its cloud infrastructure, Playtech decided to migrate to AWS in 2019. In 2020, the company began migrating its production workloads to AWS. By June 2021, CloudFront had become Playtech’s primary CDN provider for live video delivery around the world. Next, its video team began working at the application level, customizing its architecture to take advantage of the agility and performance of AWS services.

In addition to global availability and uptime, Playtech’s highest priority throughout the migration was the quality of the user experience. Its video streams need to be reliable and delivered with minimal latency so that players around the world can watch and participate in real time. “The previous system broke down too often when facing traffic spikes or anything out of the ordinary,” says Viesturs Proškins, head of video research and development for live casino at Playtech. “On AWS, we rapidly reached the state where everything was reliable, even under unpredictable pressure.” The company has greatly improved its user experience and can now support more simultaneous users than before. As a result, its clients can launch large marketing events and its infrastructure can scale efficiently to handle them.

For example, Playtech secures rights to well-known brands to use in new products and the accompanying launch events, which can cause significant spikes in traffic. And its business-to-business customers often run marketing campaigns simultaneously. Previously, Playtech had to limit the number of players in an event because its infrastructure couldn’t support these spikes with adequate video quality in all locations. Now, it can scale on AWS to deliver video streams with parity of quality across disparate geographical locations. “More people can play our games with an equally high-quality user experience during these events,” says Proškins. “Scaling is no longer an issue.” Improved scaling has also translated to cost optimization for Playtech. Before migrating, the company had to apply updates across four separately managed operating environments each with unique characteristics. Now, it has consolidated that work on a complete and seamless architecture on AWS.

In addition, Playtech built a custom edge compute solution that delivers video streams in several formats while sourcing content from multiple Regions efficiently using Lambda@Edge, a feature of CloudFront that lets users run code closer to application users, which improves performance and reduces latency. By implementing consistent, hash-based dynamic origin selection, the company can split traffic between multiple Regions and across independent Amazon Elastic Compute Cloud (Amazon EC2) Amazon EC2 Auto Scaling groups, which businesses can use to add or remove compute capacity to meet changes in demand. Each group can scale independently, improving the scalability and resiliency of the entire infrastructure. By using Lamdba@Edge container re-use, average code runtime takes only 3.7 ms to make a routing decision.

Playtech operates in 30 regulated jurisdictions and streams video from many studios around the world. Changing global regulations require the company to be able to enter new regions promptly. On AWS, Playtech can take video feeds from studios in new locations to comply with local regulations without worrying about cloud availability or reliability. “Previously, the time required to expand geographically was heavily dependent on deployment size, as the process was fully manual,” says Proškins. “Now, with everything in place on AWS, we can move into and out of a new Region in 10–15 minutes.” Instead of manually launching infrastructure and running standby servers to meet peak traffic—which wasted money and resources—the company has simplified and automated what used to be a complex deployment process using AWS CloudFormation, which speeds up cloud provisioning with infrastructure as code. Playtech uses CloudFormation templates to provision new Regions with everything they need in a simple and repeatable way. Then, the company can quickly and consistently add each newly deployed group of instances to its CloudFront distributions using Lambda@Edge consistent hash ring logic. Further, this infrastructure scales automatically to meet demand on Amazon EC2. With the simplicity and scale of its resilient architecture on AWS, Playtech can focus on adding value rather than doing maintenance. “Now, we can focus on bringing value to the business,” says Proškins. “Instead of spending time on maintenance tasks, we can build new features and use CloudFront to deliver them globally.”

Playtech’s migration to AWS is complete, and the company is already preparing to replace its public connection to AWS with a private one. This will remove the last pain points remaining from its old architecture.

Migrating to AWS has improved the availability and reliability of Playtech’s video streams while making its architecture simpler and more efficient. “It just works—that’s the expectation from users when they enter our games.” says Proškins. “And on AWS, it just works.”