Skip to main content

Kubernetes system requirements

Kubernetes version support

You need a running Kubernetes cluster, using Kubernetes version 1.16 up to 1.22. Make sure the corresponding tooling, such as kubectl is compatible with the Kubernetes version you are using.

k8s VersionWorks with Glasnostic
1.16Yes (k8s end of life)
1.17Yes (k8s end of life)
1.18Yes (k8s end of life)
1.19Yes
1.20Yes
1.21Yes
1.22Yes

Glasnostic officially supports Kubernetes deployments on these managed Kubernetes platforms:

Other standard Kubernetes cluster should work as well, please test carefully before going into production. Please contact us if you need additional support for your Kubernetes deployment.

note

Ensure other service mesh components (e.g. Istio) are not configured to inject into the same namespaces as Glasnostic to avoid conflicts. If you are already utilizing Istio see Glasnostic for Istio, which works seamlessly with any existing Istio installation.

Requirements for a Kubernets based deployment

The Glasnostic data plane is set to consume less than 200m CPU and 200MB of memory. Each sidecar injected into your workloads will need additional memory. By default it will request 20MB of memory and 5m CPU from the Kubernetes cluster. Depending on how much traffic the service is processing up to 2 CPUs and 1GB of memory may be consumed by the sidecar.

Testing and Development environments

For development or testing cluster we recommend to have at least the following sizing:

  • 4 vCPUs
  • 16 GB of RAM
  • at least 30 GB of disk space

Production environments

For production clusters we recommend to always run a multi-node cluster where each worker node has enough CPU, memory and disk capacity to support your workload. The Glasnostic data plane sidecar will consume more memory and CPU in higher traffic environments so factor this into your sizing considerations.

We recommend to run at least the following sizing for each worker node:

  • 4 vCPUs per node
  • 16 GB of RAM per worker
  • 50 GB disk per node
  • 5 Gbit networking
note

Consider all workloads planned for the cluster, the total number of expected pods, number of expected deployments and expected traffic when planning your Kubernetes cluster deployment.

Additional considerations

Before deploying Glasnostic into production make sure it's compatible with additional components in your cluster. Test especially the combination of Ingress and API Gateway solutions, and additional sidecar injection mechanisms.

We do test different combinations internally and expect no issues but you should verify with your very own setup.