Amazon EC2

Definition

Amazon Elastic Compute Cloud (EC2) is AWS’s virtual server service — it provides resizable compute capacity in the cloud. EC2 lets you launch virtual machines (instances) in minutes, choose the CPU, memory, storage, and networking configuration, and pay only for what you use. It is the foundational compute service used by approximately 64% of enterprises running workloads on AWS.

How EC2 Works

An EC2 instance is a virtual server running on AWS physical hardware. You choose:

• Instance type: CPU, memory, and network profile (e.g., m7g.large = 2 vCPU, 8 GB RAM)
• AMI (Amazon Machine Image): Operating system + software template (Amazon Linux, Ubuntu, Windows, custom)
• Region and Availability Zone: Where the instance runs
• Storage: EBS volumes (persistent) or instance store (ephemeral)
• Networking: VPC, subnet, security groups, Elastic IP

Once launched, you connect via SSH (Linux) or RDP (Windows) and run your application.

Instance Families

Graviton Processors (ARM-based)

AWS Graviton4 (M8g, C8g, R8g instances) and Graviton3 processors deliver:

• Up to 40% better price-performance vs equivalent x86 instances
• Up to 60% less energy consumption (important for Sustainability pillar)
• Support for Java, Python, Node.js, Go, .NET, and most Linux workloads

Graviton5 (M9g, released late 2025) provides 192 ARM cores per chip — double the previous generation.

Always evaluate Graviton instances first; migrate if your application supports Linux ARM64.

Pricing Models

On-Demand: Pay per hour/second, no commitment. Use for unpredictable workloads.

Savings Plans / Reserved Instances: 1 or 3-year commitment for up to 72% savings. Use for stable baseline workloads.

Spot Instances: Bid on spare AWS capacity for up to 90% savings. Instances can be reclaimed with 2-minute notice. Use for fault-tolerant batch jobs, CI/CD, and ML training.

Dedicated Hosts: Physical server dedicated to your use. Required for per-socket/per-core software licenses and some compliance requirements.

Auto Scaling

EC2 Auto Scaling automatically adjusts the number of instances based on demand:

• Target Tracking: Maintain average CPU at 60% — add/remove instances automatically
• Scheduled Scaling: Scale out before predicted traffic spikes (e.g., before business hours)
• Predictive Scaling: Uses ML to forecast traffic and scale proactively

Combine Auto Scaling with Elastic Load Balancing (ALB/NLB) for production-grade scalability.

Common Mistakes

Mistake 1: Overprovisioning instance size. Use AWS Compute Optimizer and CloudWatch metrics to right-size; most teams can reduce EC2 spend 20–40% without performance impact.

Mistake 2: Running x86 instances without evaluating Graviton. Graviton offers better price-performance for most Linux workloads — always benchmark before staying on x86.

Mistake 3: Using On-Demand for baseline workloads. Identify stable baseline usage and cover it with Savings Plans or Reserved Instances; use On-Demand or Spot only for variable demand above that baseline.

Related AWS Services

• Amazon EBS: Persistent block storage for EC2 instances
• Elastic Load Balancing: Distribute traffic across multiple EC2 instances
• EC2 Auto Scaling: Automatically scale instance count with demand
• AWS Compute Optimizer: Right-sizing recommendations based on actual usage
• Amazon VPC: Network isolation for EC2 instances

Related FactualMinds Content

• Cloud Cost Optimization Services
• FinOps Consulting
• AWS Well-Architected Review Checklist