Worker node plans
Two CPU tiers, same convention as VM plans: Shared for general workloads, CPU-Optimized with a much higher CPU commitment for steady, compute-heavy workloads (databases, CI runners, batch processing).Shared CPU
CPU-Optimized
Every worker has a 25 GB NVMe root disk for the OS, container images, logs and ephemeral storage. Persistent volumes are Raff Volumes provisioned by your PVCs, priced below.
Add-ons
You pay for exactly the storage your claims request — every GB billed is a GB you can use. Pay-as-you-go bills hourly per existing node; subscriptions commit the node set monthly, with PVC storage always billed hourly on top — see Billing model. Kubernetes itself uses about 1 GB of memory per node — prefer fewer, larger nodes over many small ones.