EC2 – Elastic Compute Cloud

Pricing

Types

Broadly catagorized into five types:

1. General Purpose

• balanced mix of compute, memory and network resources
• ideal for diverse workloads like web services, code repos; basically when workload and performance is uncertain

2. Compute Optimized

• ideal for cpu intensive tasks like machine learning, gaming, scientific modeling, etc.

3. Memory Optimized

• ideal for memory intensive tasks like processing large datasets, data analytics, databases; basically when your task requires a lot of memory

4. Storage Optimized

• designed for workloads requiring high performance for locally stored data like large databases, data warehouses and basically i/o intensive apps

5. Accelerated Computing

• uses hardware accelerators like GPUs to handle tasks efficiently like graphics processing, 3d rendering, machine learning, complex calculations, etc.

AMIs – Amazon Machine Images

AMI is pre configured template that includes OS, app server, pre-installed apps which is required to launch EC2 instances. It also contains block device mapping, permissions (which account/user can launch EC2 using this template), metadata, etc. So basically an AMI is a block-level disk image of the OS drive. AMIs can be AWS-provided, community provided, availabe in Marketplace or custom built. A single AMI can launch multiple EC2 instances:

    flowchart LR
    AMI -- launches --> EC2-1["EC2 Instance1"]
    AMI -- launches --> EC2-2["EC2 Instance2"]
  

Auto Scaling EC2

• dynamic adjusting number of EC2 instances based on load or demand
• you create auto scaling group which is basically a group or collection of EC2 instances that can scale in or out depending on the scaling rules you define