Compute Services¶

Compute refers to the processing power, CPUs, and memory that execute your code and applications in the cloud.

1. Amazon EC2 (Elastic Compute Cloud)¶

Amazon EC2 provides on-demand, scalable virtual machines (called instances). You have complete administrative control over the guest operating system, network configuration, and installed software.

Instance Naming Scheme¶

EC2 instance names follow a standardized structure (e.g., t4g.micro, m7i.large, c8a.xlarge):

       m   7   g   .   large
       │   │   │         │
       │   │   │         └─ Size (vCPU & RAM multiplier)
       │   │   └─────────── Processor (g = AWS Graviton ARM, i = Intel, a = AMD)
       │   └─────────────── Generation (7th generation)
       └─────────────────── Instance Family

Common Instance Families¶

Family	Name	Best For	Example
T	General Purpose (Burstable)	Dev/test, low-traffic web apps, microservices	`t4g.micro`, `t3.small`
M	General Purpose (Balanced)	Production web servers, app servers	`m7i.large`, `m9g.large`
C	Compute Optimized	High-performance web servers, batch processing, video encoding	`c7g.xlarge`, `c8a.xlarge`
R	Memory Optimized	In-memory caches, high-performance databases	`r7g.large`, `r8g.2xlarge`
G / P / Trn / Inf	Accelerated Computing	AI/ML model training, GPU rendering, deep learning inference	`g6.xlarge`, `p5.48xlarge`, `trn3`, `inf2`

The Power of AWS Graviton (ARM64)

Instances with a lowercase g (e.g., t4g, m7g, c7g) run on AWS's custom Graviton ARM-based processors. Graviton chips offer up to 40% better price-performance over comparable x86 instances. AWS's newest generation, Graviton5 (powering the M9g family), delivers massive improvements for modern containerized and AI workloads.

EC2 Pricing & Purchasing Options¶

graph TD
    EC2Pricing["EC2 Pricing Models"]
    OnDemand["On-Demand<br>Pay by second/hour. Zero commitment."]
    SavingsPlans["Savings Plans / Reserved<br>1 or 3 year commitment. Up to 72% off."]
    Spot["Spot Instances<br>Unused AWS capacity. Up to 90% off. Can be reclaimed."]

    EC2Pricing --> OnDemand
    EC2Pricing --> SavingsPlans
    EC2Pricing --> Spot

On-Demand: Pay strictly for running seconds. Ideal for short-term, spiky, or unpredictable workloads.
Savings Plans / Reserved Instances (RI): Commit to a steady amount of compute (measured in $/hr or specific instances) for 1 or 3 years. Delivers discounts up to 72%.
Spot Instances: Bid on spare AWS compute capacity for up to 90% discount. However, AWS can reclaim the instance with a 2-minute warning. Ideal for batch jobs, CI/CD runners, and stateless workers.

Launching and Connecting to an EC2 Instance¶

Open EC2 Console → Launch Instance.
Select an AMI (Amazon Machine Image) — e.g., Amazon Linux 2023 (AL2023) or Ubuntu 24.04 LTS.
Select an instance type (e.g., t3.micro or t4g.micro).
Configure a Key Pair (.pem format) and download it to your local machine.
Configure a Security Group allowing inbound SSH (port 22) from your IP.
Click Launch Instance.

Connecting via SSH¶

# 1. Restrict key file permissions (required by SSH clients)
chmod 400 my-aws-key.pem

# 2. Connect to the public IP or DNS
ssh -i "my-aws-key.pem" ec2-user@<public-ip-or-dns>

2. AWS Lambda (Serverless Compute)¶

AWS Lambda executes your code without requiring server provisioning, OS patching, or cluster management. You pay solely for the milliseconds your code executes.

graph LR
    Trigger["Event Trigger<br>(S3 Upload, API Request, SQS Message)"] --> Lambda["AWS Lambda Function<br>(Executes in milliseconds)"]
    Lambda --> Output["Write to DynamoDB / Send Response"]

Execution Model: Event-driven (triggered by HTTP via API Gateway, file uploads to S3, queue messages in SQS, or scheduled cron timers).
Supported Runtimes: Python, Node.js, Java, .NET, Ruby, and custom runtimes (like Go and Rust compiled to Linux binaries via provided.al2023).
Scaling: Automatically scales from zero to tens of thousands of concurrent executions.
Max Execution Timeout: 15 minutes per invocation.
Cost Savings Tip: Switch your Lambda architecture to arm64 (Graviton) in the configuration tab for a free ~20% price-performance gain!

Example Lambda Handler (Python)¶

import json

def lambda_handler(event, context):
    name = event.get("queryStringParameters", {}).get("name", "Cloud Explorer")

    return {
        "statusCode": 200,
        "headers": {"Content-Type": "application/json"},
        "body": json.dumps({"message": f"Welcome to AWS, {name}!"})
    }

3. Container Services: ECS, EKS & Fargate¶

Containers package code and all its dependencies into an immutable image (Docker).

graph TD
    Containers["Container Workloads"]
    ECS["Amazon ECS<br>(AWS-Native Container Orchestration)"]
    EKS["Amazon EKS<br>(Managed Kubernetes)"]
    Fargate["AWS Fargate<br>(Serverless Compute Engine for ECS & EKS)"]

    Containers --> ECS
    Containers --> EKS
    ECS -.->|Run on Serverless| Fargate
    EKS -.->|Run on Serverless| Fargate

Amazon ECS (Elastic Container Service): AWS's streamlined, highly integrated container orchestrator. Perfect for microservices without Kubernetes complexity.
Amazon EKS (Elastic Kubernetes Service): Fully managed Kubernetes control plane compatible with upstream Kubernetes tooling (kubectl, Helm).
AWS Fargate: Serverless compute engine for containers. With Fargate, you deploy your container tasks directly without managing any underlying EC2 virtual machines.

4. AWS Elastic Beanstalk & App Runner¶

AWS Elastic Beanstalk: A Platform-as-a-Service (PaaS) where you upload web application code (Java, .NET, PHP, Node.js, Python, Ruby, Go, Docker) and AWS manages provisioning, load balancing, auto-scaling, and health monitoring.
AWS App Runner: A modern fully managed service that takes container images or source code and deploys scalable web applications with automatic HTTPS, SSL, and load balancing in minutes.