Amazon Web Services (AWS) - Core Cheat Sheet

Updated 2026-07-16

Next Topic: AWS Cloud Services Cheat Sheet

🎯Take a practice test on this topic6 practice tests · 216 questions→

🧠Study flashcards on this topic159 cards · spaced repetition→

Amazon Web Services (AWS) is the world's most comprehensive cloud platform, offering over 200 fully featured services from 39 regions and 123 Availability Zones globally. AWS provides on-demand computing resources without upfront capital investment, transforming how organizations build, deploy, and scale applications. The platform operates on a pay-as-you-go pricing model and has expanded rapidly into generative AI with Amazon Bedrock, Amazon Nova foundation models, and Amazon Q Developer. One critical insight: AWS services are designed to work together seamlessly. Understanding how EC2, VPC, IAM, and S3 interconnect is far more valuable than knowing each service in isolation.

What This Cheat Sheet Covers

This topic spans 20 focused tables and 161 indexed concepts, 159 flashcards, 6 practice tests with 216 questions. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Compute ServicesTable 2: Storage ServicesTable 3: S3 Storage ClassesTable 4: Database ServicesTable 5: Networking and Content DeliveryTable 6: VPC Networking ComponentsTable 7: Security, Identity, and ComplianceTable 8: Management and GovernanceTable 9: Monitoring and ObservabilityTable 10: Developer Tools and CI/CDTable 11: Messaging and StreamingTable 12: Analytics and Big DataTable 13: Machine Learning and AITable 14: Auto Scaling and High AvailabilityTable 15: Cost Optimization and Pricing ModelsTable 16: Disaster Recovery and BackupTable 17: Well-Architected Framework PillarsTable 18: Global Infrastructure ConceptsTable 19: Service Limits and QuotasTable 20: Tagging and Organization

Quick IndexSubscribe to unlock

A jump-to index of every table row in this cheat sheet.

Mind MapSubscribe to unlock

An interactive map of every table and concept in this topic.

Table 1: Compute Services

AWS compute options sit on a spectrum of control. At one end EC2 hands you a whole virtual server with OS-level access and the patching that comes with it; at the other, Lambda runs your code with no servers to manage and Fargate runs containers without you provisioning instances. Pick by two questions: how much operational burden you want to keep versus hand to AWS, and how your workload is packaged, whether as a plain process, a container, or an event handler.

Service	Example	Description
EC2 (Elastic Compute Cloud)	`aws ec2 run-instances --image-id ami-0abcdef1234567890 --instance-type m8g.large`	• Virtual servers in the cloud with full OS control • supports Linux, Windows, macOS across instance families: general purpose (M8g), compute (C8g), memory (R8g), storage, accelerated • Graviton ARM-based instances cost up to 20% less than comparable x86 instances, but AMIs, binaries, and container images must be built for `arm64`
Lambda	`exports.handler = async (event) => {` `return { statusCode: 200, body: 'Hello' };` `};`	• Serverless compute that runs code in response to events • billed per request plus duration rounded up to the nearest 1 ms, auto-scales horizontally to match demand • hard ceiling of 15 minutes per invocation, so longer jobs belong in Batch, ECS, or Step Functions • SnapStart cuts init-driven cold starts to as low as sub-second on Java 11+, Python 3.12+, and .NET 8+ (not Node.js, Ruby, or container images)
ECS (Elastic Container Service)	`aws ecs create-cluster --cluster-name my-cluster`	• Fully managed Docker container orchestration • integrates natively with ALB, CloudWatch, IAM • capacity options: EC2, Managed Instances, Fargate, or ECS Anywhere • Express Mode (Nov 2025) turns a container image plus two IAM roles into a Fargate service with HTTPS, autoscaling, and a URL
EKS (Elastic Kubernetes Service)	`eksctl create cluster --name prod-cluster --region us-west-2`	• Managed Kubernetes, certified Kubernetes-conformant, so standard manifests and community tooling work unchanged • supports both EC2 and Fargate compute, Auto Mode for simplified cluster management
Fargate	`"requiresCompatibilities": ["FARGATE"]`	• Serverless compute for containers, no EC2 instances to manage • works with both ECS and EKS, right-sized and billed per task • not automatically cheaper: it wins on lightly used instances, while a well-packed EC2 fleet can cost over 20% less
Batch	`aws batch submit-job --job-name nightly-report --job-queue processing-queue`	• Fully managed batch processing at any scale • automatically provisions compute and distributes work based on the jobs in the queue • runs on EC2 and Fargate, with Spot capacity for interruption-tolerant jobs
Elastic Beanstalk	`eb init && eb create prod-env`	• Platform as a Service (PaaS) that handles deployment, scaling, load balancing, and health monitoring • supports Go, Java, .NET, Node.js, PHP, Python, Ruby, and Docker • not serverless: it provisions EC2 instances in your own account, and while Beanstalk itself adds no charge, those resources bill whether or not traffic arrives
Lightsail	`aws lightsail create-instances --instance-names my-server --availability-zone us-east-1a --blueprint-id amazon_linux_2023`	• Simplified VPS with a low, predictable monthly price • bundles compute, SSD storage, DNS, static IP, and a data transfer allowance into one plan • ideal for simple web apps and dev/test; the OS stays yours to manage

Table 2: Storage Services

AWS storage breaks down by how the data is accessed: object storage (S3) for files reached over HTTP, block storage (EBS) that attaches to a single EC2 instance like a disk, and shared file systems (EFS, FSx) that many instances can mount at once. Glacier handles cheap long-term archival, and Storage Gateway bridges your on-premises servers into the cloud. Picking the right one starts with asking how your application expects to read and write the data.

Service	Example	Description
S3 (Simple Storage Service)	`aws s3 cp file.txt s3://my-bucket/`	• Object storage with 99.999999999% (11 nines) durability, separate from its 99.99% availability target • unlimited data, ideal for backups, static websites, data lakes • 8 storage classes from Express One Zone to Glacier Deep Archive • strong read-after-write consistency in all Regions; new buckets block public access and encrypt with SSE-S3 by default

The rest of this section is part of the full cheat sheet

Table 3: S3 Storage Classes

Within S3, the storage class you choose trades storage price against retrieval cost, speed, and minimum-duration commitments. The rule of thumb: the less often you read an object, the colder and cheaper a class you can pick, from Standard for hot data, through the Infrequent Access tiers, down to the Glacier archives. The catch is that colder classes add per-GB retrieval fees, minimum billable object sizes, and minimum storage durations, so a lower per-GB rate can still raise your bill on data that is hot, small, or short-lived.

Class	Example	Description
S3 Standard	`--storage-class STANDARD`	• Default class for hot data accessed frequently • low latency, high throughput • no minimum storage duration, no retrieval fee

The rest of this section is part of the full cheat sheet

Table 4: Database Services

AWS offers a purpose-built database for almost every data shape rather than forcing everything into one engine. RDS and Aurora cover traditional relational workloads, DynamoDB handles NoSQL key-value at massive scale, Redshift is the analytics warehouse, and the specialists (Neptune for graphs, DocumentDB for MongoDB-style documents, Keyspaces for Cassandra) round out the rest. Start from the access pattern your data demands, then match it to the engine here.

Service	Example	Description
RDS (Relational Database Service)	`aws rds create-db-instance --db-instance-identifier mydb --engine postgres`	• Managed relational databases: PostgreSQL, MySQL, MariaDB, Oracle, SQL Server, Db2 • automates backups and patching • Multi-AZ adds a synchronous standby with automatic failover; read replicas are asynchronous and must be promoted manually

The rest of this section is part of the full cheat sheet

Table 1: Compute Services

Service	Example	Description
EC2 (Elastic Compute Cloud)	`aws ec2 run-instances --image-id ami-0abcdef1234567890 --instance-type m8g.large`	• Virtual servers in the cloud with full OS control • supports Linux, Windows, macOS across instance families: general purpose (M8g), compute (C8g), memory (R8g), storage, accelerated • Graviton ARM-based instances cost up to 20% less than comparable x86 instances, but AMIs, binaries, and container images must be built for `arm64`
Lambda	`exports.handler = async (event) => {` `return { statusCode: 200, body: 'Hello' };` `};`	• Serverless compute that runs code in response to events • billed per request plus duration rounded up to the nearest 1 ms, auto-scales horizontally to match demand • hard ceiling of 15 minutes per invocation, so longer jobs belong in Batch, ECS, or Step Functions • SnapStart cuts init-driven cold starts to as low as sub-second on Java 11+, Python 3.12+, and .NET 8+ (not Node.js, Ruby, or container images)
ECS (Elastic Container Service)	`aws ecs create-cluster --cluster-name my-cluster`	• Fully managed Docker container orchestration • integrates natively with ALB, CloudWatch, IAM • capacity options: EC2, Managed Instances, Fargate, or ECS Anywhere • Express Mode (Nov 2025) turns a container image plus two IAM roles into a Fargate service with HTTPS, autoscaling, and a URL
EKS (Elastic Kubernetes Service)	`eksctl create cluster --name prod-cluster --region us-west-2`	• Managed Kubernetes, certified Kubernetes-conformant, so standard manifests and community tooling work unchanged • supports both EC2 and Fargate compute, Auto Mode for simplified cluster management
Fargate	`"requiresCompatibilities": ["FARGATE"]`	• Serverless compute for containers, no EC2 instances to manage • works with both ECS and EKS, right-sized and billed per task • not automatically cheaper: it wins on lightly used instances, while a well-packed EC2 fleet can cost over 20% less
Batch	`aws batch submit-job --job-name nightly-report --job-queue processing-queue`	• Fully managed batch processing at any scale • automatically provisions compute and distributes work based on the jobs in the queue • runs on EC2 and Fargate, with Spot capacity for interruption-tolerant jobs
Elastic Beanstalk	`eb init && eb create prod-env`	• Platform as a Service (PaaS) that handles deployment, scaling, load balancing, and health monitoring • supports Go, Java, .NET, Node.js, PHP, Python, Ruby, and Docker • not serverless: it provisions EC2 instances in your own account, and while Beanstalk itself adds no charge, those resources bill whether or not traffic arrives
Lightsail	`aws lightsail create-instances --instance-names my-server --availability-zone us-east-1a --blueprint-id amazon_linux_2023`	• Simplified VPS with a low, predictable monthly price • bundles compute, SSD storage, DNS, static IP, and a data transfer allowance into one plan • ideal for simple web apps and dev/test; the OS stays yours to manage

Table 2: Storage Services

Service	Example	Description
S3 (Simple Storage Service)	`aws s3 cp file.txt s3://my-bucket/`	• Object storage with 99.999999999% (11 nines) durability, separate from its 99.99% availability target • unlimited data, ideal for backups, static websites, data lakes • 8 storage classes from Express One Zone to Glacier Deep Archive • strong read-after-write consistency in all Regions; new buckets block public access and encrypt with SSE-S3 by default

Table 3: S3 Storage Classes

Class	Example	Description
S3 Standard	`--storage-class STANDARD`	• Default class for hot data accessed frequently • low latency, high throughput • no minimum storage duration, no retrieval fee

Table 4: Database Services

Service	Example	Description
RDS (Relational Database Service)	`aws rds create-db-instance --db-instance-identifier mydb --engine postgres`	• Managed relational databases: PostgreSQL, MySQL, MariaDB, Oracle, SQL Server, Db2 • automates backups and patching • Multi-AZ adds a synchronous standby with automatic failover; read replicas are asynchronous and must be promoted manually