Nine Entertainment Achieves 98% Cost Savings with Amazon ElastiCache Serverless
Nine Entertainment, a prominent Australian media company behind streaming services like 9Now and Stan, has reported extraordinary cost savings by transitioning its caching infrastructure to Amazon ElastiCache Serverless for Valkey. According to an AWS Database Blog post from May 19, 2026, the company achieved a staggering 98% reduction in caching costs, alongside enhanced scalability and the elimination of manual operational overhead during high-demand periods. This success story challenges the common perception that serverless offerings are inherently more expensive than provisioned alternatives, especially for high-throughput workloads.
The core challenge for Nine Entertainment was that the caching infrastructure for their streaming platform APIs, which serve hundreds of thousands of concurrent users, was costing more than the underlying compute platform. Their previous Amazon ElastiCache for Redis OSS deployment, while providing sub-millisecond latency, required constant monitoring and manual adjustment of node counts to handle the variable load of streaming services. This led to either over-provisioning and increased costs during quiet periods or capacity shortages during sudden traffic surges.
Amazon ElastiCache Serverless for Valkey addressed these issues by automatically scaling compute, memory, and network resources both vertically and horizontally to precisely match application demand. This 'pay-only-for-what-you-consume' model proved to be up to 30% cheaper than node-based ElastiCache clusters for applications with fluctuating traffic patterns. Beyond cost savings, the migration freed engineering teams from maintaining scaling runbooks and monitoring clusters before major events, allowing them to focus on innovation. The combination of payload compression and the serverless architecture provided both cost efficiency and operational peace of mind, demonstrating a powerful case for serverless adoption in dynamic environments.
Read original source