Live Streaming UEFA Euro 2024 for 1.2 Million Concurrent Devices Using AWS with Deutsche Telekom

Germany was hosting the Euro 2024 organized by the Union of European Football Associations (UEFA), and integrated telecom company Deutsche Telekom (DT) was one of the main sponsors. Given the worldwide excitement for the sports event, DT expected a manifold increase in viewers on its streaming product, OneTV. The global company wanted to help viewers watch every match with virtually no downtime or outages.

OneTV was built on Amazon Web Services (AWS), and DT reviewed the product architecture, identified potential choke points, and enhanced the architecture for scalability, resilience, and low latency. The company streamed 51 live matches over the 30-day event on its enhanced platform, providing a smooth viewing experience to millions of enthusiasts, and even reduced its per-subscriber cost.

OneTV runs on Amazon Elastic Kubernetes Service (Amazon EKS)—a managed Kubernetes service to run Kubernetes seamlessly on AWS and in on-premises data centers—and uses several other AWS and open-source services. In 2023, the team migrated from Elasticsearch to Amazon OpenSearch Service—which lets organizations securely unlock near real-time search, monitoring, and analysis of business and operational data. By removing the need to manage the operational overhead of self-hosted Elasticsearch, the company gained a more scalable and configurable solution that simplified search and monitoring capabilities for OneTV. To further increase scalability and resilience, it migrated to Amazon DocumentDB (with MongoDB compatibility)—a fully managed native JavaScript Object Notation (JSON) document database.

To assess scalability requirements, the team analyzed traffic patterns from the Champions League event—which is similar to UEFA Euro—looking at prime or nonprime match timings, weekday and weekend matches, and whether the country where a match was being streamed was playing in that match. It then came up with an algorithm for scaling on match days. The team tested the scaling strategy in different countries, increasing the capability to scale up by as much as 10 times.

“We knew we would need to scale up by 5 or 10 times at any moment,” says Pradeep Verma, DevOps lead at DTDL. “So, we introduced our own custom scaler and added a scheduler.” The scaler helped to scale OneTV in a country that was playing in a particular live match, and the DTDL team could schedule it to scale up and down for different countries at different times, depending on the matches being streamed. The team also scaled components such as databases and Amazon EKS in different ways to enhance performance.

The team used Amazon DocumentDB to build resilience. The database cached a list of the URLs that a customer had visited previously on OneTV through their service provider. In case of a service provider outage, OneTV’s resilience model could read the cache from Amazon DocumentDB to fail over to another URL and provide uninterrupted viewing to the customer.

The DTDL team also worked alongside the AWS Enterprise Support to create a disaster recovery site in 30 days. Frankfurt was the primary AWS Region for OneTV, and the Ireland AWS Region became the secondary site in case of major outages. “We set up automations and configurations in both Regions so we could scale up or down from any Region anytime,” says Verma.

The DTDL team used AWS Countdown—which lets organizations optimize business-critical events, product launches, migrations, and modernizations on AWS—to work alongside the AWS account team throughout the project lifecycle. The teams collaborated to assess cloud operational readiness, identify and mitigate risks, plan capacity, and conduct operational reviews. This helped the DTDL team to achieve its objectives for OneTV within its timeline and budget threshold, reducing per-subscriber cost by about 48 percent.

DTDL enlisted AWS Support in the countdown to the event and during the event as well. “We raised support cases during match screenings, and the AWS team was very proactive in getting them resolved,” says Srivastava. Because the product worked so well, only 10 support cases were raised in the 30-day event period.

The DTDL team screened the matches in its offices in different countries, and the operations team monitored the service 24/7 for any issues. The system effortlessly handled a six-times increase in traffic as API requests climbed from 2,500 to 9,300 per second—the highest in OneTV history.

OneTV streamed 51 live matches—26 million hours of content—over the 30-day event to 1.2 million concurrent devices in Europe, with virtually no interruptions, downtime, or latency issues. “With the collaboration between our teams and the support from AWS, we reached a near-perfect system design that works flawlessly,” says Srivastava. “This experience has helped me see the scale and impact of this collaboration, and I’d like to apply this learning to other products.”