AWS - Deep Understanding of EC2 Instance Types

Amazon EC2 instances offer a variety of features designed to enhance performance, flexibility, and cost-effectiveness for different workloads.

AWS has the following naming convention

Ex: m5.xlarge

m: instance Class

5: generation ( AWS takes care of improving hardware then generation will change)

xlarge: size within the instance class.

Here are some Instance Features.

Instance Types:

1. General Purpose

2. Compute Optimized

3. Memory Optimized

4. Accelerated Computing

5. Storage Optimized

General Purpose

• General-purpose instances provide a balance of

  • Compute

  • Memory

  • Networking

• Examples of workloads include web servers and code repositories.

• Below are the general-purpose EC2 Instances that can be used.

  

Compute Optimized

• Compute Optimized Instance type is ideal for high-performance processors.

• Below are the categories that are well-suited

  • Batch Processing Workloads

  • Media transcoding

  • High-performance web servers

  • High performace computing(HPC)

  • Scientific modeling & Machine Learning

  • Dedicated gaming servers

• Below are the EC2 Instances that can be used.

Memory Optimized

• These are designed to deliver fast performance for workloads that process large data sets in memory

• Below are the categories that are most used:

  • High-performance databases, such as Relational and Non-relational databases

    • eg, MySQL, PostgreSQL, MongoDB, etc

  • distributed web-scale in-memory caches.

  • Real-time big data analytics.

  • Electronic Design Automation

  • SAP HANA Workloads

  • Machine Learning Model Training

• Below are the EC2 instances that can be used for this Instance type

Accelerated Computing

• These are used for hardware acceleration or co-processors

• Below are the use cases

  • Floating point number calculations

  • graphics processing

  • Data pattern matching

  • Computer Vision, like image and video analysis

  • Financial Modeling, like high-frequency trading algorithms.

• These are the EC2 instances that can be used

Storage Optimized

• Storage-optimized instances are designed for workloads that require high, sequential read and write access to large datasets on local storage.

• These instances are optimized to deliver tens of thousands of low-latency, random I/O operations per second(IOPS) to applications.

Below are some use cases

• High-performance databases such as NoSQL databases, Time Series databases, and in-memory databases

• Distributed file systems like HDFS ( Hadoop Distributed File System)

• Data warehousing

• Log and data processing

• Data Lakes and many more.

Processor Options:

• x86-based processors (Intel and AMD)

• Arm-based AWS Graviton processors

Burstable Performance

• Instances like T3 provide a baseline level of CPU performance with the ability to burst above the baseline.

EBS Optimization

• Dedicated throughput between Amazon EC2 and Amazon EBS

Enhanced Networking

• Higher packet per second (PPS) performance, lower inter-instance latencies, and lower network jitter

Auto Recovery

• Automatically recover instances in case of system impairments

  Placement Groups

  • Logical grouping of instances to optimize for networking performance

Elastic Fabric Adapter (EFA)

• A network interface for Amazon EC2 instances that enables low-latency, high-throughput networking

Instance Store

• High-performance, temporary block-level storage

Elastic Inference

• Attach low-cost GPU-powered acceleration to EC2 instances

AWS Nitro System

• A combination of dedicated hardware and a lightweight hypervisor for enhanced security and performance

Hibernation

• Preserve the in-memory state of an instance for quick restarts

When selecting instance types and features, consider your specific workload requirements, performance needs, and cost constraints. Always refer to the official AWS documentation for the most up-to-date information on instance features and availability across regions.