A Brief Guide To And Comparison Of Amazon EC2 Instance Types

CloudHealth Tech Staff
Published:
Jun. 27, 2019
4 minute read

Amazon Web Services frequently releases new Amazon EC2 Instance types or upgrades to existing instance types. Here we look at the current generation of AmazonEC2 Instance types, and compare how they perform against previous generations in order to illustrate how much more cost-efficient the cloud has become in the past decade.

In August 2006, Amazon Web Services launched its Elastic Compute Cloud (EC2) service in beta mode exclusively to existing customers. Originally available in the only AWS Region around at the time - US East-1 - the virtual servers had the equivalent of a 1.7 GHz Xeon processor, 1.75 GB of memory, 160 GB of storage, and 250 Mb/second of network bandwidth. They cost $0.10 per hour to run.

As more instance sizes were introduced, the original virtual servers became known as “small” instances and a new term was introduced to describe instances’ processing capabilities - the “EC2 Compute Unit”. EC2 Compute Units represent the number of compute units allocated to each instance multiplied by the number of virtual cores, making it easier to compare capabilities of different Amazon EC2 Instance types.

 

Different Amazon EC2 Instance types became a “thing” in 2008 when AWS launched its first compute-optimized instances. The new Amazon EC2 Instance types had proportionally more CPU power than memory, making them more suitable (and cost-effective) for CPU-intensive applications. By the end of the year, AWS customers had a choice of five different instance types:






Amazon EC2 Instance Types - 2008

Instance

EC2 Compute Units

Memory (GB)

Storage (GB)

Price per Hour

Small

1

1.7

160

$0.10

Large

4

7.5

850

$0.40

X-Large

8

15.0

1690

$0.80

High-CPU Medium

5

1.7

350

$0.20

High-CPU XL

20

7.0

1690

$0.80

More Amazon EC2 Instance Types Added

Over the years, the range of instance sizes and Amazon EC2 Instance types has grown significantly. AWS users currently have a choice of up to 135 different EC2 instances depending on their location (not all instance configurations are available in every Region). The range of instances is broken down into five Amazon EC2 Instance types to make navigating the choice easier:

General Purpose 

This type includes the “main” or “happy medium” M-type EC2 instances, burstable T-type EC2 instances, and recently released A1 EC2 Instances powered by arm-based AWS Graviton processors.

Compute Optimized

This type consists of C-type EC2 instances similar to the “High-CPU” instances launched in 2008, only with more compute power than previously and better overall performance.

GPU “Accelerated Computing” Instances

Accelerated Computing instances also have three different prefixes - “P”, “G” and “F”. Generally these AWS EC2 instance types are suitable for high graphics workloads or deep learning workloads.

Memory Optimized

This category of Amazon EC2 Instance types has gone through some significant changes recently to cater for extremely high memory requirements and includes R-type, X-type and Z1d-type EC2 instances.

Storage Optimized

Storage optimized EC2 instances features instances for HDD storage (H-type), high IOPS throughput (I-type), and an extremely high ratio of storage compared to CPU and RAM (D-type).

As each of the Amazon EC2 Instance types is upgraded, the number following its prefix is changed to indicate which generation it belongs to. For example, the latest generation of M-type General Purpose EC2 instances (M5a) has more bandwidth than its predecessor and improved network performance. It also has slightly faster processors than the M4 or M5 instance type.

It’s worth keeping up-to-date with any news about EC2 releases and upgrading instances to the latest generation whenever appropriate. As a rule, not only do new generation EC2 instances deliver better performance than their predecessors, they can also be cheaper as well. For example, the better performing m5a.2xlarge EC2 instance costs $0.344 per hour (US East-1 Region), compared to $0.384 per hour for an m5.2xlarge instance, and $0.400 per hour for an m4.2xlarge instance.

How Prices and Performance Compare over a Decade

Even though there were only five Amazon EC2 Instance types a decade ago, it’s not possible to make a direct EC2 comparison because, as well as now measuring memory capacity in Gibibytes (GiBs) rather than Gigabytes (GBs), most EC2 instances have Elastic Block Storage (EBS) attached by default. Nonetheless, we have attempted to produce the closest possible EC2 comparison table below:






Amazon EC2 Instance Types - 2019

Instance

EC2 Compute Units

Memory (GiB)

Storage (GB)

Price per Hour

t2.small

variable

2.0

EBS

$0.023

m5a.large

4

8.0

EBS

$0.086

m5a.xlarge

8

16.0

EBS

$0.172

c5.large

9

4.0

EBS

$0.085

c5.xlarge

17

8.0

EBS

$0.170

As you’ll see by comparing the two tables, the difference in prices is significant. Plus, the prices displayed are On Demand prices that don’t take into account discounts for Reserved Instances that could reduce the 2019 EC2 prices by up to 70% - not forgetting that in all circumstances you’ll receive a better level of performance than you would have in 2008.

Managing Amazon EC2 Instance Types

Knowing which types of EC2 instances are most suitable for your workloads, and then managing different Amazon EC2 Instance types sounds complicated, but it doesn’t have to be. With a cloud management platform such as CloudHealth, you receive cost and utilization recommendations for EC2 instances.

The CloudHealth platform will notify you when instance utilization meets the threshold to justify a Reserved Instance purchase. Taking advantage of CloudHealth’s capabilities can save you a considerable amount of time and money; and if you’d like the opportunity to evaluate our cloud management platform in your own environment, contact us today to request a free trial of CloudHealth.