Challenges in Managing Dedicated Amazon EKS Clusters
Provisioning dedicated Amazon EKS clusters for each testing environment proved to be a resource-intensive practice for Deloitte. The process consumed up to 45 minutes per cluster, delaying critical quality assurance (QA) tasks. QA engineers required isolated environments for specific testing scenarios, but the reliance on platform teams for setup introduced operational bottlenecks.
Each testing environment incurred significant infrastructure duplication, such as separate load balancers, DNS configurations, and monitoring agents. Additionally, managing multiple AWS IAM roles and Kubernetes RBAC settings increased complexity, straining team resources and slowing development cycles. The cumulative result was higher costs, extended provisioning times, and reduced team autonomy.
Understanding the Cost Implications
The reliance on multiple dedicated clusters led to an increase in AWS infrastructure expenses. Each cluster required its own set of resources, including Application Load Balancers and Amazon Route 53 records, which added to the overall operational budget. This approach made scaling testing environments financially unsustainable over time.
The platform team's involvement in cluster provisioning further escalated labor costs. Time that could have been spent on core development activities was instead diverted to repetitive infrastructure tasks. This dual cost-both financial and human-highlighted the need for a more efficient solution.
Implementing Amazon EKS with vCluster
To address these challenges, Deloitte adopted a solution combining Amazon EKS with vCluster. Amazon EKS provided the foundational compute and networking infrastructure, while vCluster enabled the creation of lightweight, virtual Kubernetes clusters. These virtual clusters functioned as fully independent environments, eliminating the need for multiple dedicated Amazon EKS clusters.
This approach streamlined resource utilization by allowing multiple virtual clusters to share the same underlying EKS infrastructure. As a result, Deloitte reduced the operational overhead associated with managing distinct environments. The platform team could now focus on strategic tasks, freeing up valuable time and resources.
Improvements in Provisioning Times
By transitioning to vCluster, Deloitte achieved 89% faster provisioning for testing environments. What previously took 30-45 minutes could now be accomplished in just a fraction of that time. This improvement enabled QA engineers to quickly set up isolated environments without relying heavily on platform teams.
The reduction in setup time had a direct impact on the overall development cycle. Teams could iterate faster, identify issues more quickly, and deploy fixes without delays. This enhanced agility translated into a more efficient testing process and accelerated delivery timelines.
Operational and Financial Benefits
The adoption of vCluster not only improved provisioning times but also delivered substantial cost savings. By sharing resources across virtual clusters, Deloitte reduced expenses related to load balancers, DNS entries, and other duplicated infrastructure components. The consolidated setup also minimized maintenance efforts, lowering ongoing operational costs.
Moreover, the streamlined process enhanced team autonomy. QA engineers could independently manage their environments, reducing dependencies on platform teams. This autonomy further boosted productivity and allowed Deloitte to allocate resources more strategically, achieving a balance between performance and cost-efficiency.