How we helped achieve a $100K/year saving in cloud costs by switching cloud providers

Optimizing cloud infrastructure costs without sacrificing performance is a persistent challenge in today's rapidly evolving tech landscape. In this blog, we share how we helped a client save approximately $100,000 per year by transitioning their containerized workloads from Amazon ECS (Elastic Container Service) to Google Cloud Run.

Why We Considered Leaving AWS ECS

Our client operates several containerized applications using AWS ECS with Fargate. A significant limitation of ECS is its inability to scale services down to zero instances when they're idle, meaning the services continuously incur costs regardless of actual traffic volume. This was especially costly for our client, whose workloads naturally experience large fluctuations in traffic—busy at certain times and nearly idle at others.

Google Cloud Run’s Auto-scaling Advantage

We explored Google Cloud Run because of its unique serverless nature, particularly its ability to scale workloads completely down to zero when not in use. This capability directly aligned infrastructure costs with actual usage, eliminating unnecessary expenses during low-traffic periods.

Navigating the Migration

Assessing and Preparing for the Transition

We began by closely evaluating each ECS workload to determine suitability for Cloud Run. Fortunately, most of the client's services were stateless, making them ideal candidates for serverless container environments.

Addressing Cold Starts

One challenge with Cloud Run's scale-to-zero feature is managing cold start latency—the brief delay experienced when a service spins up from zero instances. While noticeable, we confirmed these cold starts remained well within acceptable limits for our client's operations. We also proactively optimized container images and implemented strategies like periodic warm-up requests to further reduce these delays.

Executing the Migration

Optimizing Containers for Cloud Run

We carefully migrated each service, streamlining containers to enhance their startup performance and overall efficiency on Cloud Run.

Seamless DNS Transition

After deploying the services to Cloud Run, we smoothly updated DNS configurations, ensuring minimal downtime and no disruption for end-users.

Comprehensive Testing

Extensive testing was a crucial step in this process. We conducted load testing, security assessments, and thorough performance monitoring under diverse conditions to ensure stability and reliability.

Results and Insights

Switching from AWS ECS to Google Cloud Run resulted in significant annual savings of about $100,000 for our client. Cloud Run’s scale-to-zero capability made the biggest difference by aligning infrastructure costs precisely with usage patterns. Despite managing cold starts, the performance and reliability of our client’s applications improved noticeably.

This experience highlights the importance of regularly reviewing your cloud strategy and making informed decisions to achieve better cost efficiencies and performance.

Note: This blog describes our specific experiences and outcomes. Individual results may vary based on different use cases and configurations.