Decode AWS EC2 Instance Names: A Complete Guide to Families, Generations, and Specs

AWS EC2 offers a wide variety of instance types, and their seemingly complex names follow a clear set of rules. Understanding these rules enables rapid selection of the most suitable instance for a given application.

Instance Type Naming Structure

EC2 instance names consist of four parts:

<family>.<generation>.<(metal)>.<size>

. For example, t3.medium, m5.large, r5d.xlarge, and c6gn.16xlarge.

Breakdown of Each Component

Instance Family : The first letter indicates the primary optimization direction.

T : Burstable Performance (e.g., web servers, dev environments).

M : General Purpose.

C : Compute Optimized.

R : Memory Optimized.

X : Extreme Memory.

I : Storage Optimized.

D : Dense-storage.

G : GPU Instances.

P : General GPU Compute.

F : FPGA Instances.

Inf : Inference Instances.

Trn : Training Instances.

Instance Generation : The number following the family denotes the generation; higher numbers mean newer hardware with better price‑performance (e.g., m5 is newer than m4).

Metal Indicator : Optional suffixes provide extra hardware details. metal: Bare‑metal instance (no hypervisor). d: NVMe local SSD storage. n: Network‑optimized.

Instance Size : The final part describes the resource size, using descriptors ( small, medium, large, xlarge) or multipliers ( 2xlarge, 16xlarge). Larger sizes provide more vCPUs, memory, and network capacity.

Example Decoding

t3.medium

: t = burstable family, 3 = third generation, medium = moderate size. m5.large: m = general purpose, 5 = fifth generation, large = larger size. r5d.xlarge: r = memory‑optimized, 5 = fifth generation, d = NVMe SSD, xlarge = extra‑large. c6gn.16xlarge: c = compute‑optimized, 6 = sixth generation, g = Graviton ARM processor, n = network‑optimized, 16xlarge = 16× extra‑large.

Special Identifiers

Processor Architecture a: AMD CPUs (e.g., m5a.large). g: AWS Graviton (ARM) CPUs (e.g., m6g.large).

None: defaults to Intel CPUs.

Storage d: NVMe local SSD. i: Variant of local SSD used by some families.

Network Optimization n: Enhanced networking (higher bandwidth, lower latency).

Compute Performance z: High single‑thread performance.

Bare‑Metal metal: Bare‑metal instance (e.g., m5.metal).

Accelerated Computing p: General GPU. g: GPU for graphics‑intensive workloads. inf: AWS Inferentia chip for ML inference. trn: AWS Trainium chip for ML training.

High‑Memory z: Instances with very large memory footprints (e.g., x2idn.16xlarge).

Conclusion

By mastering the EC2 instance naming conventions, you can quickly infer an instance's capabilities and appropriate use cases, enabling better resource optimization and cost control on AWS.