Cluster configuration guidelines

Guidelines for cluster configuration:

• Control plane
  • 3 control plane nodes
  • Vault will not be scheduled on the control plane in accordance with general Kubernetes community recommendations
  • Vault does not place unusual demands on the control plane relative to other general workloads
• Worker nodes
  • A minimum of 3 worker nodes is needed to enable failure tolerance
  • The internal database that Vault uses is optimized for SSD drives
  • CPU and storage performance requirements will depend on the customer's exact use profile
    • Use Vault metrics to understand the load
• Infra nodes
  • 3 infrastructure nodes
  • Traffic routing and monitoring pods will be hosted on the infra nodes

Recommended cluster configuration

Stakater recommends this cluster configuration:

• Two Kubernetes clusters to host four Vault clusters for sake of HA for two sites
• Four Vault clusters (two prod and two non-prod) is the minimum setup to enable High Availability (HA) and Disaster Recovery:
  • Two Vault clusters running on one prod Kubernetes cluster:
    • One primary active prod Vault
    • One secondary passive non-prod Vault (DR)
  • Two Vault clusters running on one non-prod Kubernetes cluster:
    • One primary active non-prod Vault
    • One secondary passive prod Vault (DR)
• Each Kubernetes cluster will have three control plane nodes and six worker/infra nodes
• Each node (control plane/worker/infra) will be 4 vCPUs x 16 GB RAM x 100 GB disk with minimum 3000 IOPS
• Each Vault cluster will be three nodes that will run on three worker/infra nodes
• So, one Kubernetes cluster will be 36 vCPUs x 144 GB RAM x 900 GB disk
• Each cluster will have two separate sharded ingresses