Server & VM Sizing Calculator
Turn workload signals into an estimated starting-point capacity — vCPU, RAM, storage, instance count, HA/DR topology and approximate cloud mappings.
About this tool
What it checks
- Estimated vCPU and RAM for your workload signals
- Storage capacity with sensible headroom
- Recommended instance count for availability
- A suggested HA/DR topology (single, multi-node, multi-AZ)
- Approximate mappings to AWS, Azure, GCP and OCI general-purpose families
How it works
- 1You enter workload signals (users, requests, data, growth) — everything runs in your browser.
- 2A documented sizing model turns those signals into a starting-point capacity estimate.
- 3The result includes redundancy and an approximate cloud instance mapping per provider.
- 4You validate the estimate with load testing before committing — it is a starting point, not a guarantee.
Example output: For a mid-size web app: ~8 vCPU / 32 GB across 2–3 instances behind a load balancer, ~500 GB storage, mapped to comparable AWS/Azure/GCP/OCI general-purpose instances.
Frequently asked questions
It is a deterministic engineering estimate, not guaranteed capacity. It maps your workload signals (concurrency, requests/sec, CPU/memory intensity, app type, growth, HA/DR) to a starting-point capacity using documented coefficients and safety margins. Always validate with load testing before production.
It is fully documented and shown under 'Sizing methodology & assumptions'. In short: effective load drives raw vCPU (1 vCPU ≈ 50 req/sec at medium CPU intensity), a safety margin and environment factor are applied, RAM is derived from vCPU × memory intensity × app-type bias, then capacity is scaled out into instances. Growth is projected 12 months forward.
No. They are approximate mappings to general-purpose families (AWS m5, Azure Dsv5, GCP n2, OCI E4.Flex). Per-vCPU performance, storage and network differ between providers — treat them as starting points, not equivalence guarantees.
Yes. HA sets a minimum of two instances across availability zones and adds a load-balanced topology; DR adds a warm-standby region at roughly 40% of production capacity as a starting point. Both are shown as explicit trade-offs.
Use it to get to a credible starting point quickly. For production sign-off, validate with load testing against your real workload — the tool says this on every result.
Related services
Found this tool useful? Help keep it free.
Our tools have no sign-up, no ads and no paywall. A one-time contribution of any amount covers servers and funds new tools.