Leaseweb Introduces Kubernetes Vertical Pod Autoscaler (VPA)

We’re excited to announce an improvement to our Managed Kubernetes service: Leaseweb is now using Vertical Pod Autoscaler (VPA) to optimize the control plane components of your clusters. This enhancement focuses specifically on the core Kubernetes components that Leaseweb manages, such as etcd, kube-scheduler, and kube-api-server, ensuring they scale efficiently as your cluster grows.

What’s Being Enhanced?

The VPA implementation specifically targets the following control plane components managed by Leaseweb:

etcd (distributed key-value store)
kube-scheduler
kube-api-server
Other core control plane services

Key Benefits of Vertical Pod Autoscaler

VPA is a controller / operator that runs in each Kubernetes cluster and monitors a set of pods for Memory and CPU usage. According to the actual usage of these pods reported by the metrics-server and out-of-memory events, Vertical Pod Autoscaler will give recommendations on how to set resource requests and resources limits for your pods and optionally configure the resource with those recommendations.

How VPA optimizes Leaseweb’s Control Plane

Implementing the insights generated from Vertical Pod Autoscaler, will result in the following benefits:

Automatically meet scaling needs: Leaseweb’s Managed Kubernetes solution includes Vertical Pod Autoscaler, which automatically sets optimal resource requests and limits for your containers. This eliminates the need for manual provisioning, ensuring your applications have the right resources to perform at their best.

Smarter scheduling: Having closer-to-reality resource requests will make sure we avoid scheduling essential components of your cluster on a node that is already full or near-limit.

Denser Kubernetes cluster: Having the resource requests really close to actual usage will result in denser cluster and avoid over-provisioning.

Ease anomaly detection: When a pod starts deviating from nominal usage and reaches the resource limit. The Pod will be killed for another to takeover.
This can happen if a new image version of the pod leaks memory, or some actions causes the pod to consume more memory than usual.

More stable platform. By ensuring sure all pods that run in a cluster are well-managed in terms of memory and CPU usage, we avoid surprises and that will result in a more stable platform.

Understanding VPA’s Role in Control Plane Management

Key highlights:

Monitoring CPU and memory usage of control plane components
Analyzing resource utilization patterns
Automatically adjusting resource allocations for optimal performance

Important Note for Customers

This VPA implementation is specifically for the Kubernetes control plane components managed by Leaseweb. It does not affect:

Your application workloads
Customer-managed nodes
Custom resources in your namespace

Getting Started

You can experience these control plane optimizations by:

Ordering Leaseweb Managed Kubernetes Beta Cluster
Upgrading your existing cluster (contact Leaseweb support for details)

Leaseweb Enhances Control Plane Management with Vertical Pod Autoscaler (VPA)