AWS EC2 — Learning Roadmap
Estimated Time to Job-Ready
5-7 weeks of consistent learning (2-3 hours/day) — EC2 itself is conceptually simple (a VM), but real proficiency means understanding instance family selection, pricing models, and how EC2 sits inside a VPC, security groups, and auto-scaling — not just launching an instance.
Phase 1: Foundation (Week 1-2)
•Instance families and sizing: general-purpose (m-series), compute-optimized (c-series), memory-optimized (r-series), and when each is actually the right choice for a workload
•Launching an instance via the console and CLI, connecting via SSH/Session Manager, and understanding the difference between stopping, terminating, and rebooting
•AMIs (Amazon Machine Images): using a public AMI vs. building a custom one, and why a "golden AMI" pipeline matters for consistent deployments
•Security groups (stateful, instance-level firewall) vs. NACLs (stateless, subnet-level) — a genuinely common early point of confusion
Checkpoint: can you explain why a security group rule allowing inbound traffic on port 443 doesn't need a matching outbound rule for the response, while a NACL does?
Phase 2: Pricing Models and Storage (Week 2-4)
•On-Demand vs. Reserved Instances vs. Savings Plans vs. Spot Instances — the real tradeoff each makes between cost and commitment/interruption risk
•EBS volume types (gp3, io2, st1) and matching volume type to actual I/O pattern, not defaulting to one type everywhere
•Instance store vs. EBS — ephemeral vs. persistent storage, and why losing instance store data on stop/terminate surprises people who assumed all EC2 storage persists
•Elastic IPs vs. dynamic public IPs, and why a dynamic IP changing on stop/start breaks naive DNS assumptions
Checkpoint: given a genuinely fault-tolerant, interruption-tolerant batch workload, can you justify choosing Spot over On-Demand with a real cost comparison, not just "Spot is cheaper"?
Phase 3: Auto Scaling and Production Patterns (Week 4-6)
•Auto Scaling Groups: launch templates, scaling policies (target tracking, step scaling), and health checks that actually replace unhealthy instances
•Placement groups (cluster, spread, partition) and when each matters for real workloads
•User data scripts for instance bootstrapping, and why baking more into the AMI beats a longer, more fragile boot-time script
•Load balancer integration (ALB/NLB) with an Auto Scaling Group as the standard production pattern
Checkpoint: can you explain why a target-tracking scaling policy is generally preferred over manually-tuned step scaling for most workloads?
Phase 4: Cost, Security, and Interview Readiness (Week 6-7)
•Right-sizing based on actual CloudWatch utilization data, not initial provisioning guesses
•IAM instance profiles/roles for granting an EC2 instance AWS permissions without embedding long-lived credentials
•Patch management strategy (Systems Manager Patch Manager, or golden-AMI rotation) for fleets at scale
•Review this course's Interview Q&A material, particularly instance family selection and Spot vs. Reserved tradeoff questions
Common Pitfalls Specific to EC2
•Assuming instance store data survives a stop/start — it doesn't; only EBS-backed volumes persist across a stop, and instance store is wiped on both stop and terminate
•Treating security groups and NACLs as interchangeable — security groups are stateful and instance-level; NACLs are stateless and subnet-level, requiring explicit rules for both directions
•Hardcoding AWS credentials into an instance instead of using an IAM instance profile — a long-lived embedded credential is a real security liability an instance profile avoids entirely
•Choosing an instance family by habit rather than actual workload profile — a memory-bound workload on a compute-optimized instance (or vice versa) wastes money without the performance benefit you're actually paying for
Getting Your First EC2-Heavy Role
1.Portfolio: a project demonstrating a real Auto Scaling Group behind a load balancer, with a genuine scaling event triggered and documented (not just configured)
2.Resume: be specific — "reduced EC2 spend 30% by right-sizing over-provisioned m5.2xlarge instances to m5.xlarge based on 30 days of CloudWatch utilization data" is far stronger than "experience with AWS EC2"
3.Know the pricing models cold: interviewers commonly probe whether you can justify a pricing model choice with real reasoning, not just recite the four options
4.Certifications, if pursuing one: EC2 is heavily featured in AWS Solutions Architect Associate and SysOps Administrator — see this course's own Certification Guide for current format and pricing