FreeWheel Migrates Key Ad Tech Platforms to AWS, Delivering a Globally Scalable Cloud Transformation

Acquired by Comcast in 2014, FreeWheel provides comprehensive managed advertising solutions for publishers, advertisers, and media companies. It operated by using a colocation model globally, with multiple physical locations that combined hardware with networking services. The infrastructure was costly to scale to meet the company’s more than 30 percent growth in its advertising transaction volumes year over year. FreeWheel was looking for a solution to streamline and cut costs for its flagship product, the Monetization Rights Management (MRM) solution. MRM is a complete ad sales and management technology that supports three critical workloads for customers.

Its primary workload involves FreeWheel’s ad servers, which make instantaneous decisions about which ads to show online. The company required a manageable way to handle spikes in traffic that occur around major live events such as the Super Bowl. The ad servers generate copious logs that clients collect and analyze to optimize their ad campaigns, resulting in a big data workload. Also, FreeWheel had to run its web application, including user interfaces for clients’ ad operation service teams. “AWS was at the leading edge, and we immediately launched compute resources and high-level services,” says Alex Ng, senior vice president of global operations at FreeWheel. “We could solve some of our growing pains from volume spikes, which were not easy to solve with on-prem infrastructure.”

In 2017, FreeWheel launched its Cloud First initiative and began its migration to AWS, starting with compute-heavy data-processing workloads. To run its large Apache Hadoop cluster, FreeWheel spun up multiple instances of Amazon Elastic Compute Cloud (Amazon EC2), which provides secure and resizable compute capacity for virtually any workload. As the company quickly grew, it built in elasticity through the use of Amazon Elastic Kubernetes Service (Amazon EKS), a managed service to run Kubernetes in the cloud and on-premises data centers. “We’ve been able to keep our team size steady while supporting production that is growing in multiple magnitudes,” says Ng. “It’s a big win.”

The company incorporated additional native AWS services. It migrated core databases to Amazon Aurora, designed for unparalleled high performance and availability at a global scale with full MySQL and PostgreSQL compatibility. “Using Amazon Aurora, our databases are globally synchronized,” says Xiaoli Liu, senior director of the FreeWheel site reliability engineering team. “There is very low latency for updates, changes, or usage.”

FreeWheel found that it could implement other applications seamlessly alongside AWS solutions. In 2021, the company acquired Beeswax, whose bidder-as-a-service solutions had been built natively on AWS. Furthermore, FreeWheel’s in-house monitoring solution, which clients use to collect and analyze logs to optimize their ad campaigns, worked well on AWS.

FreeWheel moved its forecasting module and MRM user interface to AWS and further moved toward a serverless architecture based on AWS Lambda, a serverless, event-driven compute service that lets organizations run code for virtually any type of application or backend service without provisioning or managing servers. “We spend more time talking to clients to understand their business requirements instead of focusing on scaling our infrastructure,” says Ng. “We are seeing a very positive change in how our team operates.”

By migrating its on-premises workloads to AWS, FreeWheel scaled to accommodate capacity requirements for the Tokyo Olympic Games in 15 minutes, a process that would have taken 6 months using its legacy MRM infrastructure. For the 2022 FIFA World Cup, the company alerted AWS that it anticipated six million concurrent users so that AWS could adequately prepare network load balancers. “For each match, we spun up thousands of ad servers in a blink,” says Liu. “And when the match concluded, those newly deployed ad servers automatically disappeared to save costs when the peak was gone. That’s the elasticity and flexibility we gain from AWS.” To further improve performance while optimizing cost, FreeWheel converted compute instances for its monitoring, forecasting, and other critical components to AWS Graviton processors, which are based upon Arm architecture and are designed by AWS to deliver the best price performance for cloud workloads running in Amazon EC2. To further reduce costs, FreeWheel implemented Savings Plans, a flexible pricing model that helps reduce the monthly bill by up to 72 percent compared with On-Demand prices, in exchange for a spend commitment. All told, FreeWheel saves up to 20 percent in compute costs using AWS Graviton processors.