Skip to content

Compute: EC2, ASG, and Load Balancing

Overview

Provision and scale Amazon Elastic Compute Cloud (EC2) safely: Amazon Machine Images (AMIs), launch templates, Auto Scaling groups (ASGs), placement, Elastic IP (EIP) awareness, and Elastic Load Balancing with Application Load Balancer (ALB) and Network Load Balancer (NLB).

EC2 is virtual servers in your Virtual Private Cloud (VPC). Production systems rarely run a single instance: you use launch templates, ASGs across Availability Zones (AZs), and a load balancer for health checks and traffic distribution. This module connects compute to the networking patterns from Module 3.

Cost

Running EC2, ALB/NLB, NAT, and unattached Elastic IPs incurs charges. Prefer t3.micro / Free Tier types, --dry-run, stop or terminate promptly, and delete load balancers and target groups. Spot is cheaper but interruptible — not for this first lab unless you accept interruption.

This is a core tutorial in Module 4 · Compute of the REBASH Academy AWS for Cloud & DevOps Engineers series — written for Cloud, DevOps, Platform, and SRE engineers.

Prerequisites

Learning Objectives

By the end of this tutorial, you will be able to:

  • Choose instance family, AMI, launch template, and subnet for a workload
  • Design a multi-AZ Auto Scaling group with desired/min/max and health replacement
  • Contrast placement group strategies and Elastic IP anti-patterns
  • Choose Application, Network, or Gateway Load Balancer for a traffic pattern
  • Sketch ALB → target group → ASG with safe lab teardown

Architecture

This topic’s control points and relationships are shown below.

AWS compute

Theory

What it is

Amazon Elastic Compute Cloud (EC2) provides resizable virtual machines in your Virtual Private Cloud (VPC). Production systems rarely run a single instance: you encode launch configuration in Amazon Machine Images (AMIs) and launch templates, keep capacity with Auto Scaling groups (ASGs), optionally control placement, and put Elastic Load Balancing in front for health checks and traffic distribution. This module connects compute to the networking patterns from Module 3.

Why it matters

Continuous Delivery needs golden AMIs or controlled bootstrap plus ASG rolling refreshes — not snowflake servers. Platform teams place ALBs, target groups, and health checks in front so SRE can drain connections during deploys. DevSecOps uses least-privilege instance profiles and Session Manager instead of open SSH to the internet. Wrong balancer choice or Elastic IP habits on fleets create cost, scaling, and security debt.

How it works

  1. Choose an AMI and instance type.
  2. Apps in private subnets; load balancer in public subnets.
  3. Encode AMI, type, security groups, IAM instance profile, user data, and volumes in a launch template; the ASG pins a version.
  4. Target group + listener (for example 443); health checks drive ASG replacement.
  5. Scaling policies adjust desired capacity; three-tier path: ALB → app ASG → database (Module 6).

Keep user data short and idempotent; prefer baked images for production.

Concept deep dive

  • EC2 — Virtual servers with chosen CPU, memory, networking, and storage. You pay while instances run (and for some attached resources when stopped). Pick families for general purpose, compute, memory, or storage-optimised work; start labs on Free Tier–eligible types.
  • AMIs — Bootable templates that include the root volume snapshot, launch permissions, and block-device mapping. Use Amazon-published images for learning; use organisation golden AMIs (patched, hardened, agents preinstalled) for production fleets.
  • Launch templates — Versioned launch specifications: AMI, instance type, key pair (optional), security groups, IAM instance profile, user data, and Elastic Block Store mappings. Prefer launch templates over legacy launch configurations. ASGs and RunInstances both consume them.
  • Auto Scaling — An ASG maintains desired capacity between min and max. It launches and terminates instances across subnets/AZs, replaces instances that fail ELB or EC2 health checks, and scales on CloudWatch metrics, schedules, or predictive policies. Rolling instance refresh updates the fleet to a new launch-template version.
  • Placement groups — Cluster packs instances close together for low latency (same AZ, risk of correlated failure). Spread places instances on distinct hardware to reduce blast radius (limits on count). Partition spreads across racks/partitions for large distributed datastores. Default (no placement group) is fine for most web tiers.
  • Elastic IP (EIP) — A static public IPv4 address you can remap. Useful for rare fixed endpoints or migration windows. Costly when unattached; an anti-pattern on every ASG member — prefer load balancers or DNS that tracks changing private/public IPs.
  • Load balancers — Application Load Balancer (ALB) operates at Layer 7 (HTTP/HTTPS): host/path routing, TLS termination, container and microservice targets. Network Load Balancer (NLB) operates at Layer 4 (TCP/UDP/TLS): ultra-low latency, static IPs, PrivateLink. Gateway Load Balancer (GWLB) steers traffic through third-party virtual appliances (firewalls, inspection) using GENEVE — use when you insert inline network functions, not for ordinary web apps.

Key concepts and comparisons

Concept Guidance
Instance profile IAM role on the instance — preferred credential source
Desired / min / max ASG capacity bounds and scaling headroom
Target group Health checks + routing destination for ELB
ALB vs NLB vs GWLB L7 apps → ALB; L4/static IP → NLB; appliances → GWLB
Connection draining Deregistration delay for in-flight requests
IMDSv2 Require tokens for instance metadata (SSRF defence)
Spot vs On-Demand Cost versus interruption risk

Common pitfalls

  • Calling a single instance in one AZ “production”
  • Public IP plus wide SSH security group instead of Session Manager
  • Using launch configurations (legacy) instead of launch templates
  • ALB without health checks, or checks that hit the wrong path
  • Forgetting to delete ALB/NLB after labs (hourly charge)
  • Elastic IPs left unassociated or glued to every ASG member
  • Choosing GWLB when you only needed an ALB for HTTP routing

Hands-on Lab

Create a workspace for this tutorial.

mkdir -p ~/rebash-aws/module-04 && cd ~/rebash-aws/module-04

Focus: inventory EC2/ASG/ELB with describe APIs

Step 1 – Describe compute landscape

aws sts get-caller-identity
aws ec2 describe-instances --query 'Reservations[].Instances[].{Id:InstanceId,State:State.Name,Type:InstanceType}' --output table
aws elbv2 describe-load-balancers --query 'LoadBalancers[].{Name:LoadBalancerName,Type:Type}' --output table 2>/dev/null || true
aws autoscaling describe-auto-scaling-groups --query 'AutoScalingGroups[].{Name:AutoScalingGroupName,Desired:DesiredCapacity}' --output table 2>/dev/null || true

Step 2 – Optional create skipped by default

echo "Skipped create by default — describe-only is enough"
echo "If you launch an instance: aws ec2 terminate-instances --instance-ids <id>"

Final step – Cleanup note

# COST WARNING: prefer describe/list APIs. Destroy anything you create.
# Keep ~/rebash-aws/ for later tutorials

Validation

  • Lab commands run under ~/rebash-aws/module-04/
  • You can explain each Theory section in your own words
  • You used modern tooling where it applies to this topic
  • You can describe one production failure mode for this topic

Code Walkthrough

Production practice for Compute: EC2, ASG, and Load Balancing always combines:

  1. Inspect before you change (status, plan, logs, dry-run)
  2. Prefer reversible, documented changes (Git, IaC, drop-ins, version pins)
  3. Capture evidence (command output, pipeline logs) for handovers
  4. Prefer current tools and APIs over legacy shortcuts
  5. Least privilege — escalate credentials only when required

Keep runbooks short enough to follow under pressure. Automate checks; keep humans for judgement.

Security Considerations

  • Treat credentials and tokens for aws as privileged — never commit them
  • Prefer short-lived auth (OIDC, roles, SSO) over long-lived keys
  • Validate blast radius before apply/deploy/delete operations
  • Restrict who can approve production changes
  • Collect audit logs; limit who can read sensitive traces

Common Mistakes

Calling a single instance in one AZ “production”

Validate assumptions against the Theory section and official docs before changing production.

Public IP plus wide SSH security group instead of Session Manager

Lab shortcuts (open security groups, admin roles, skip approvals) must not ship unchanged.

Changing production without a rollback path

Always know how to revert (previous artefact, prior release, state rollback, DNS failback).

Best Practices

  • Encode Compute: EC2, ASG, and Load Balancing changes as code and review them in pull requests
  • Pin versions (images, modules, actions, provider plugins)
  • Separate environments with clear promotion gates
  • Alert on symptoms with runbooks attached
  • Destroy lab resources; tag everything with owner and expiry where possible

Troubleshooting

Symptom Likely cause Fix
Auth / permission denied Wrong identity, policy, or scope Check caller identity, roles, and least-privilege policies
Timeout / no route Network, DNS, security group, or endpoint Trace path, DNS, and allow-lists before retrying
Drift / unexpected plan Manual change or wrong state/workspace Reconcile desired vs actual; avoid click-ops on managed resources
Pipeline/job red Flaky step, cache, or missing secret Read failing step logs; bisect recent workflow/config changes
Cost spike Idle load balancer, NAT, oversized compute Inventory billable resources; stop/delete labs promptly

Summary

Compute: EC2, ASG, and Load Balancing is essential for Cloud and DevOps engineers working with aws. Practise the lab until the inspection and change path is muscle memory, then continue the track.

Interview Questions

  1. ASG desired/min/max — what do they mean?
  2. ALB versus NLB versus CLB?
  3. Instance unhealthy behind ALB — checks?
  4. Implications of stopping versus terminating?
  5. How do launch templates improve consistency?

Sample answer — question 2

Check instance state, status checks, and load balancer target health before resizing.

Sample answer — question 4

Use IMDSv2, least-privilege instance roles, and terminate lab instances promptly.

References