Proxmox VE Cluster Sizer and HA Failover Calculator
Size your Proxmox VE cluster: set node count and per-node hardware, then check vCPU overcommit ratios, per-node RAM headroom, and High-Availability (HA) N-1 failover capacity.
Cluster & Workload Parameters
Minimum 3 nodes recommended for HA Quorum.
ZFS Pool & Storage Specs
VM & LXC Workload Planned
Cluster Capacity Utilization Live Telemetry
Node Memory Allocation Breakdown (Per Node)
Engineering Recommendations
- Cluster is operating within healthy CPU overcommit parameters (< 3:1 ratio).
- ZFS ARC is adequately sized for 8 TB raw storage.
- High-Availability (HA) failover headroom is verified for N-1 node tolerance.
How to read N-1 failover headroom
A Proxmox VE cluster earns its complexity only if it survives a node going down. The N-1 test asks one question: if a single host fails, can the remaining nodes run every guest that was on the cluster? This tool sums the RAM your VMs, ZFS ARC, and PVE services actually claim, then compares that against the capacity of N minus one nodes. If the total exceeds what the survivors can hold, high availability will try to restart guests that have nowhere to fit, and some stay down. Size for N-1 from the start, because adding RAM across a live cluster is the expensive way to learn this.
What the vCPU overcommit ratio means
Virtual CPUs are time-shared, so a cluster can assign more vCPUs than it has physical cores. The ratio is total assigned vCPUs divided by total physical cores. Up to roughly 3:1 is comfortable for mixed homelab and small-production workloads; 3-5:1 is a moderate gamble that shows up as scheduling latency under load; above 5:1, CPU-bound guests contend badly. Cores, not clock speed, buy consolidation headroom. For the memory side of the same exercise see Proxmox VE hardware requirements, and model pool capacity separately with the ZFS pool capacity calculator.