Preparing Buycycle's Infrastructure for Unexpected Traffic Spikes

The challenge

A TV campaign brought an unexpected surge of visitors to Buycycle, a marketplace for pre-owned bikes. Its infrastructure used a mix of Amazon EC2, which provides virtual computers, and EKS, AWS's service for running applications managed by Kubernetes. The traffic exposed weaknesses in how workloads were distributed and how quickly capacity could respond.

Autoscaling is the process of adding or removing computing capacity as demand changes. If the rules, resources or available capacity are not set up for a sudden spike, the response can arrive too slowly. At first, the monitoring available to the team did not give enough detail to pinpoint the problem.

What we did

Das Meta simplified Kubernetes tooling that affected scaling and revisited the distribution of workloads across EC2 and EKS. We added health checks, alarms, logs and tracing: ways to check whether services respond, record what happens and follow a request through the system. We also worked with the development team on application performance.

What the team learned

A campaign can produce a sharper traffic pattern than gradual organic growth. The case highlights the value of testing sudden spikes from multiple sources and checking how overload affects health checks and input/output operations. Read more about cloud infrastructure management.