fundamentals
AWS for DevOps Engineers: The Services That Actually Matter
Beginner
12 minutes
October 2026
CloudOpsGuide Team
AWS for DevOps Engineers: The Services That Actually Matter
AWS has 200+ services, but DevOps work concentrates on about a dozen. Learn these well and you can run real production infrastructure; the rest you can pick up when a project demands it.
Table of Contents
- Compute: where code runs
- Storage & data
- Networking: where incidents live
- Identity & security
- Operations
- How to practice without a scary bill
Compute: where code runs
- EC2 — virtual machines. Understand instance types, AMIs, user-data bootstrapping, and security groups. Still the baseline everything else is compared against.
- ECS/EKS — containers on AWS's own orchestrator vs. managed Kubernetes. ECS is simpler and deeply AWS-integrated; EKS buys you the Kubernetes ecosystem and portability.
- Lambda — event-driven functions. Great for glue code, cron jobs, and spiky workloads; bad for long-running or latency-critical paths.
Storage & data
- S3 — object storage. Learn buckets, versioning, lifecycle rules, and especially bucket policies — public-bucket misconfiguration is a classic breach vector.
- EBS/EFS — block volumes for EC2 vs. shared filesystem.
- RDS — managed databases. Know Multi-AZ vs. read replicas, and why automated snapshots matter.
Networking: where incidents live
- VPC — subnets, route tables, NAT gateways. Public vs. private subnet placement is the single most important security decision in most architectures.
- Security groups — stateful firewalls on resources. Default-deny, then open minimally.
- ALB/NLB — load balancers. ALB speaks HTTP (paths, hostnames); NLB speaks TCP at massive throughput.
- Route 53 — DNS, health checks, failover routing.
Identity & security
- IAM — the service you must actually understand. Roles over users, least privilege, and never long-lived access keys in code. Instance roles and OIDC federation for CI are the modern pattern.
- Secrets Manager / SSM Parameter Store — where credentials live instead of environment files.
Operations
- CloudWatch — metrics, logs, alarms. Not glamorous, but it's the observability floor.
- CloudTrail — audit log of every API call. Turn it on, ship it somewhere safe, hope you never need it.
- Systems Manager (SSM) — Session Manager gives shell access without SSH keys or open port 22. Use it.
How to practice without a scary bill
Set a billing alarm first ($5 threshold). Use the free tier, tag everything you create, and tear down with Terraform when done. A VPC + EC2 + ALB lab costs cents per hour and teaches more than any certification video.
Related Articles
- Essential Linux Commands Every DevOps Engineer Needs
- Ansible for Beginners: Automate Server Configuration
Last Updated: October 2026 Author: CloudOpsGuide Team Difficulty: Beginner Estimated Reading Time: 12 minutes