LearnGlossary

What are AWS instance families?

Updated 26 October 2026 · Published by Zoorik

AWS instance families are sets of EC2 instance types that differ only in size. An EC2 instance is an AWS server, usually a virtual machine (VM). Most families keep the same memory per virtual CPU (vCPU) at every size, except the burstable T families.

AWS instance family vs type

The instance type is the full name, such as c7gn.xlarge. An instance family in AWS is the part before the period. AWS’s example has four parts:

  • c is the compute optimized series.
  • 7 is the generation.
  • gn holds the options: g for Arm-based Graviton processors, n for network and EBS-optimized.
  • xlarge is the size.

Three EC2 instance families at one size

At the large size, each has 2 vCPUs but different memory:

  • c7i.large, compute optimized: 4 GiB
  • m7i.large, general purpose: 8 GiB
  • r7i.large, memory optimized: 16 GiB
Same 2 vCPUs, but 4, 8 or 16 GiB of memory AWS's sizes, drawn as three bars on one shared memory scale from 0 to 16 GiB, with labels at 0, 4 GiB, 8 GiB and 16 GiB. Each bar has 2 segments, one per vCPU. c7i.large, compute optimized, 2 vCPUs, 4 GiB: its bar ends at 4 GiB, and a bracket under one segment reads 2 GiB per vCPU. m7i.large, general purpose, 2 vCPUs, 8 GiB: its bar ends at 8 GiB, and a bracket under one segment reads 4 GiB per vCPU. r7i.large, memory optimized, 2 vCPUs, 16 GiB: its bar ends at 16 GiB, and a bracket under one segment reads 8 GiB per vCPU. Same 2 vCPUs, but 4, 8 or 16 GiB of memory memory04 GiB8 GiB16 GiBc7i.largeCompute optimized2 vCPUs, 4 GiB2 GiB per vCPUm7i.largeGeneral purpose2 vCPUs, 8 GiB4 GiB per vCPUr7i.largeMemory optimized2 vCPUs, 16 GiB8 GiB per vCPU
AWS’s sizes, drawn to scale: each segment is one vCPU and the memory that comes with it, 2 GiB on c7i.large, 4 GiB on m7i.large and 8 GiB on r7i.large.

Why it matters for your cloud cost

If a family doesn’t match your pods’ requests, one resource runs out first. Part of the other sits idle on every node.

Say each pod requests 0.5 vCPU and 4 GiB. This illustration ignores what a node keeps for the system.

  • An m7i.xlarge has 4 vCPUs and 16 GiB. Memory fills after 4 pods, so 2 vCPUs sit idle.
  • An r7i.xlarge has 4 vCPUs and 32 GiB. It fits 8 pods exactly.

So 8 pods need two m7i.xlarge nodes or one r7i.xlarge. Compare the price per pod before you change a node group’s instance type.

The free assessment works out what your clusters and disks could save.

Find your own number

One line in your cloud console. One PDF report with your estimated saving at list price.

Estimate your saving

Talk to the builders

Thirty minutes with the people who build Coral and Amoeba.

Book a demo (opens in a new tab)

Or write to founders@zoorik.com