AWS M3 Vs. M4 Vs. M5 Instances: A Side-By-Side Comparison

With Amazon EC2 M instances, you get balanced virtual CPU, memory, and networking resources, ideal for a wide range of workloads. The M instance family includes more than 14 instances, including M3, M4, and M5.

In this comparison, we highlight the differences between M3 and M4 instances and share the best use cases for each. We’ll also share a tool to help you pick the right instance type and size for specific use cases, based on factors such as AWS service, resource types, and pricing.

What Are Amazon EC2 M3 Instances?

Amazon M3 instances are a part of the General Purpose Amazon EC2 instances category. These are previous-generation instances along with M2 and M1. Because of this, M3 instances do not support the latest processors, Enhanced Networking, or AVX-512 instructions.

But AWS recently made M3 instances compatible with its Nitro System. That means you can now use the platform’s lightweight hypervisor and dedicated hardware with these previous-generation instances. So you can now run M3 instances on next-generation hardware without worrying about the service life of the older-generation servers that have powered the instances.

In addition, the AWS Nitro System support for M3 instances comes with security, scalability, and price-performance improvements.

There are four types of M3 instances, some of which are EBS-optimized, and all of which use 64-bit architecture processors. Also, as an M3 instance size increases, so does its network bandwidth, which ranges from moderate to high.

What Are Amazon EC2 M4 Instances?

Like M3 instances, Amazon EC2 M4 instances offer steady compute, memory, and network resources for a wide variety of use cases and workloads. However, M4 instances support Enhanced Networking and EBS-optimized storage by default and at no additional cost.

Intel’s Xeon E5-2676 v3 processors (Haswell) power M4 instances and are specifically optimized for Amazon EC2 usage. The processors run at a base speed of 2.4 GHz but can reach 3.0 GHz with Intel Turbo Boost. The processors support AVX2, which can significantly improve performance for applications using Intel SSE or Intel AVX.

M4 instances are ideal for small to mid-size enterprise applications, including cluster computing, mid-size databases, and caching fleets. The instances are also available as On-Demand, Reserved Instances (RI), or Spot Instances.

What Are Amazon EC2 M5 Instances?

M5 instances are the fifth generation of the M family of EC2 instances. When AWS launched them in 2017, it announced three new instance sizes (m5.24xlarge, m5.8xlarge, m5.16xlarge) and support for up to 318 GiB memory and 96 vCPUs per instance.

This new generation instances also support up to 25 Gbps of network capacity with Enhanced Networking and EBS (slower block level storage but persists data) or NVMe-based SSD (faster local storage for m5d instances but doesn’t persist data).

You also get to choose between five M5 instance types:

Wondering if M5 or C5 instances are the best for your application? See our Amazon M5 vs. C5 instances comparison here.

Or, use CloudZero Advisor to get custom recommendations on the right instance type and size for your use case.

AWS M3 Vs. M4 Vs. M5 Instances: Comparison Table

	Amazon M3 Instances	Amazon M4 Instances	Amazon M5 Instances
Description	Third-generation of the M instance family	Fourth-generation Amazon EC2 M instances	Fifth-generation Amazon EC2 M instances
Specialty	Deliver balanced compute, memory, and network bandwidth for a lot of different small workloads	Provide balanced compute, memory, and network capacity for mid-size applications	Offer balanced compute, memory, and network bandwidth for high-performance requirements
Processor	Previous-generation 64-bit architecture processor	2.4 GHz Intel Xeon Scalable processor	3.1 GHz Intel Scalable processor with Advanced Vector Extension (AVX-512) instructions
Max number of cores per instance	4	32	48
Max memory	30 GiB	256 GiB	384 GiB
Instance storage	160 GB maximum local – No EBS support	EBS-optimized and NVMe-based SSDs available	EBS-optimized and NVMe-based SSDs available
Network performance	Moderate to high	High for smaller instance sizes, and up to 25 Gbps for larger ones	Up to 10 Gbps for smaller instance sizes, and up to 25 Gbps for larger ones
Dedicated EBS bandwidth	EBS only available for m3.xlarge and m3.2xlarge instances	450 Mbps to 10 Gbps depending on instance size	Up to 4.75 Gbps for smaller instance sizes, and up to 19 Gbps for larger ones
Best use cases	Small databases, web servers, and business applications	Mid-size databases, cluster computing, testing environments, and gaming servers	High-performance workloads, including machine learning, analytics, and distributed gaming servers

How To Choose The Right Instances For Your Applications

Now here’s the thing. Choosing the right instance types and sizes for your project can help you achieve optimal technical performance and minimal financial cost to maximize your return on investment.

But even experienced engineers know that configuring the right instance types and sizes for a new or fast-changing computing project can be challenging. The result: trial-and-error, requiring frequent stops to change instances, and more downtime, which is bad for business and your SLA commitments.

With CloudZero Advisor, you can minimize all that uncertainty. With this tool, you can find the best instance based on several factors, like AWS Region, service, pricing, and resource type, as well as instance type and size. It is free and easy to use. Use CloudZero Advisor for free here.