VM & Server Sizing Calculator

Work out how many physical hosts a fleet of VMs needs, allowing for CPU and RAM overcommit plus N+x spare capacity. Calculates in your browser.

Hosts required (with spares)

-

Hosts for the workload-
Limited by CPU-
Limited by RAM-
Total vCPU / RAM-

Show the math

Work out the hosts CPU needs and the hosts RAM needs, then take the larger and add spares:

hosts = max(⌈vCPU ÷ (cores × ratio)⌉, ⌈RAM ÷ host RAM⌉) + spares

Overcommit lets several vCPUs share a core and (with care) RAM be oversubscribed, but push it too far and performance suffers. The N+x spares keep capacity during a host failure or maintenance. Calculated entirely in your browser.

What this does

A VM sizing calculator works out how many physical hosts a fleet of VMs needs, allowing for CPU and RAM overcommit plus N+x spare capacity.

How to use it

  1. Enter total vCPU and RAM needed.
  2. Enter host capacity and overcommit ratios.
  3. Add N+x spares.
  4. Read the host count.

How it works

CPU and RAM are sized separately — divide each demand by what a host provides, round up, and take the larger. Overcommit raises density; spares keep VMs running when a host fails.

Understanding your result

CPU and RAM are sized separately and you take the larger host count, because one resource usually runs out first, sizing on the wrong one leaves you short. Overcommit raises density but pushes contention, so be conservative on production, and always keep the N+x spare capacity so a single host failure does not take VMs down with it.

Example

A fleet needing 200 vCPUs at a 4:1 ratio on 32-core hosts needs about 2 hosts for CPU, plus a spare.

Sources & methodology

Last updated .

Frequently asked questions

How many hosts do my VMs need?

Add up the vCPUs and RAM your VMs need, divide each by what one host provides (cores × the vCPU-per-core ratio for CPU, and host RAM for memory), round each up, and take the larger of the two. CPU and RAM are sized separately because either can be the limit.

What is overcommit?

Overcommit lets virtual resources exceed physical ones, several vCPUs share a physical core, and RAM can be modestly oversubscribed. It raises density but, pushed too far, hurts performance, so set ratios you are comfortable running.

Why add spare hosts?

N+1 (or N+x) spare capacity keeps your VMs running when a host fails or is taken down for maintenance, so the cluster is not at 100% with no slack.