#CaseStudy

Kubernetes Infrastructure Optimization and AWS Cost Reduction

We conducted an audit of EKS clusters, eliminated causes of instability and resource overuse. As a result, we reduced infrastructure costs by 40% and virtually eliminated service downtime.

The client uses multiple Kubernetes clusters (EKS) on AWS to run product and internal services.

The main issues were high infrastructure costs and frequent service outages, directly impacting business financial metrics.

Сontext

Such architecture prevented safe scaling of the product and carried a high risk of downtime and data loss.

Increase the stability of service operations

Eliminate the causes of frequent restarts and performance degradations

Reduce costs on cloud infrastructure

Solution

Disabled VPA (keeping data collection mode)

For dev environments, we switched node types to t3, reducing their cost by about 50%*

Configured node consolidation policies in Karpenter to remove underutilized instances

Set up spot instances for GitHub Actions runners

We conducted an infrastructure audit and implemented a series of optimizations:

Reconfigured the Descheduler, raising the allowable node utilization to 70%

Configured requests and limits for all services

Set up HPA (Horizontal Pod Autoscaler) for services with variable load

Set the minimum number of replicas in production services to 3

Results

40% reduction in infrastructure costs

Savings achieved by autoscaling during night hours and weekends

More predictable resource planning

Virtually complete elimination of service crashes and restarts

We will analyze your current architecture, identify bottlenecks, and propose an optimal improvement plan.

Submit a request and we will contact you within one business day.

Get a free infrastructure audit and recommendations for its development.