Groundcover automates Kubernetes sizing. Engineers set the bounds.

Groundcover has launched Autonomous Infrastructure in beta, which continuously adjusts CPU and memory requests for Kubernetes Deployments based on production data. This technology, acquired from Wand in September, utilizes existing data from Groundcover’s eBPF sensor to inform resource allocation decisions while providing engineers with visibility into how those decisions impact infrastructure costs and workload performance.

CEO and co-founder Shahar Azulay emphasizes the importance of striking a balance between efficiency and stability in the environment. The goal is not just to reduce resource requests but to ensure workloads receive the necessary resources without overpaying for unused capacity. Groundcover aims to create a continuous feedback loop that uses real production telemetry to identify when workloads are over- or under-provisioned and make safe adjustments.

While there are various methods to right-size Kubernetes workloads, such as scheduled rightsizing, in-place resizing, and automated recommendation tools, the challenge lies in finding the right balance between cost efficiency and application stability. Workloads are constantly changing, yet resource configurations often remain static after initial deployment.

This can lead to over-provisioning, which, though safer, incurs unnecessary costs, or under-provisioning, which can cause performance issues and outages. Groundcover leverages production data from its eBPF sensor to recommend CPU and memory requests for Kubernetes Deployments, analyzing workload behavior over time to anticipate capacity requirements and prepare for expected changes in demand.

This approach helps ensure workloads have the necessary resources when demand arises without permanently allocating for peak consumption, potentially leading to significant cost savings.