m4.large

Analysis

The m4.large is a general purpose instance in the m4 series with 2 vCPUs and 8 GiB of memory on Intel Xeon E5-2676 v3 (Haswell) (up to 2.4 GHz), which makes it balanced at 4 GiB of memory per vCPU.

Running it in N. Virginia instead of São Paulo saves $43.07 a month per instance: $0.1000/hour against $0.1590/hour, a 59% spread.

Reserving it for a year with no upfront payment cuts the Linux rate to $0.0620/hour, which beats on demand once it runs more than 62% of the time; three years brings it to $0.0432/hour, 57% below on demand. Windows Server roughly doubles the bill at $0.1920/hour, because the licence is charged per vCPU.

On demand it costs $0.1000/hour in US East (N. Virginia), $73.00 a month if left running, which puts it mid-table for general purpose types at the 35th percentile per vCPU ($0.0500 per vCPU-hour) and the 35th percentile per GiB ($0.0125 per GiB-hour).

The newer m5.large on Intel Xeon Platinum 8175 is 4.0% cheaper at $0.0960/hour, which makes the m4.large hard to justify for new deployments.

Looking beyond its own line, 14 of 41 current types with 2 vCPUs and similar memory are cheaper. The cheapest of those is the t4g.large on AWS Graviton2 at $0.0672/hour. It is burstable, so it only suits workloads that do not run the CPU flat out.