How to Choose the Right AWS EC2 Instance: A Complete Guide

EC2 (Elastic Compute Cloud) is a very popular service from AWS (Amazon Web Services). It's like renting a computer from them. They've been offering this service since 2006.

I'm going to talk about what you should think about when you want to choose an EC2 instance.

An EC2 instance is kind of like your own computer. When you want to buy a computer, you might think about:

• How fast it can work?
• How much memory it has?
• How much storage it has?

These are the same things to consider when choosing an EC2 instance. But remember, with EC2, you're not buying the computer, you're renting it from AWS.

Each EC2 instance has:

• CPU (how fast it can work)
• Memory
• Storage (some have it attached, some use separate network storage called EBS)

Let me explain this using an analogy. Think of an EC2 instance like your desk at work:

• Your brain is like the CPU (the thinking power)
• The surface of your desk is the memory (where you keep things for quick access)
• The desk drawer is the storage (where you keep things for later)

What is compute?

It's about doing many things at once, like multitasking. Your brain can multitask to some extent. For example, you can talk on the phone while taking notes. But you can't write two different letters at the same time or talk on the phone while taking notes and reading a book.

Why? Because our brains are like a CPU with just one core. To do more at once, you need more cores in the CPU.

Memory and storage are similar in that they store data. But they work differently:

• Memory is fast but temporary, like things on your desk's surface.
• Storage is slower but permanent, like things in your desk drawer.

So, when choosing an EC2 instance, consider how much multitasking power (compute), quick-access memory, and storage space you need. It's a bit like choosing the right tools for your work desk.

How to choose the appropriate EC2 instance

When picking an EC2 Instance, you have three main things to consider: CPU, memory, and sometimes storage. Remember, storage is often chosen separately using EBS volumes unless it's a storage-optimized instance that comes with its own attached storage.

When you choose an instance type, you're essentially choosing the best price for what matters most for your work. This could be CPU/GPU power, memory, or storage space.

There are five types of AWS instances:

1. General Purpose: Good for tasks needing a bit of everything like compute power, memory, and networking. Best for web servers.
2. Compute Optimized: Great for jobs needing powerful processors. Best value for CPU cores. Useful for intense tasks like scientific modeling or gaming.
3. Accelerated Computing: Ideal for heavy tasks needing a lot of computing power. It uses GPUs for machine learning or high-performance computing.
4. Storage Optimized: Perfect for jobs needing high read/write speeds for lots of data, known as high IOPS. Great for NoSQL databases like Elasticsearch or Cassandra.
5. Memory Optimized: Designed for tasks needing lots of RAM. Best value for memory. Ideal for in-memory databases or SQL databases.

Each type has its best uses:

• General Purpose is good for balanced tasks.
• Compute Optimized is for powerful processors.
• Accelerated Computing is for extra demanding jobs.
• Storage Optimized is for high-speed data work but doesn't save data if the instance stops.
• Memory Optimized is for tasks needing lots of RAM like databases or caches.

Anatomy of an EC2 Instance Name

Have you encountered EC2 instance names such as t2.nano, r6a.large, or i3en.6xlarge and wondered about the significance of their letters and numbers?

Let's break down a complex instance name like c4gn.xlarge for a better understanding.

Instance Family:

The initial letter in the instance name represents its family, which aligns with specific instance types—general purpose, compute-optimized, accelerated computing, storage-optimized, or memory-optimized. Familiarity with these families tends to grow as you work more within AWS.

For instance, i3en.6xlarge belongs to the "i" family, indicating it's a storage-optimized instance.

Instance Generation:

The succeeding number signifies the instance generation. Higher numbers denote more recent generations. Opting for the latest generation often offers better hardware, leading to a lower cost per performance unit compared to older versions.

In the example of i3en.6xlarge, it's a third-generation instance.

Special Features:

Additional letters following the generation denote specific features of the instance. For example, 'e' signifies extra capacity, like RAM or storage, while 'n' indicates network optimization, implying high bandwidth.

There are various special feature characters and corresponding capabilities like 'a' for AMD processors, 'g' for AWS Graviton processors, 'i' for Intel processors, 'd' for instance store volumes, 'b' for block storage optimization, and 'z' for high frequency. These additional features come at a cost, so they should be chosen only if necessary.

Instance Size:

The size designation after the dot comprises a number and letters denoting the size, ranging from nano to xlarge. For xlarge instances, the number indicates multiples of the base xlarge size.

The figure after the dot functions as a multiplier for compute power (vCPUs), memory, and, in some cases, storage capacity. For instance, an i3en.xlarge instance may have 4 vCPUs, 32 GiB memory, and 2500 GB storage, while an i3en.6xlarge is six times larger, featuring 24 vCPUs, 192 GiB memory, and 15,000 GB storage.

Selecting Your Instance:

Choosing an EC2 instance involves several logical steps:

1. Select instance type based on workload requirements.
2. Choose an instance family aligned with the workload's technical recommendations.
3. Consider additional features only if necessary, as they come at an added cost.
4. Determine the appropriate instance size through iterative testing, aiming for a balance between utilization and headroom to handle peak demand without performance issues.

Ultimately, the goal is to choose an instance type that optimizes cost-effectiveness for your workload's key performance metric. Finding the right instance size often involves an iterative process, starting small and scaling up as needed.

Sign up for my newsletter to receive more valuable content. If you enjoyed reading my article, please show your support by giving it a round of applause ?? and sharing it on your LinkedIn profile.

Feel free to connect with me across various social media channels! Join me on LinkedIn, YouTube, Twitter, GitHub, and Upwork. Your support means a lot.