Free · no account needed

AWS SQS — From First Principles to Production

12 of 61chapters written in full

The rest are published as outlines — the curriculum and structure are settled, the prose is still being written.

A complete course on Amazon Simple Queue Service, written for engineers who need to design, operate and defend SQS-based systems — not just call the API.

It assumes no prior knowledge of messaging systems. It ends with you able to size a consumer fleet with arithmetic, diagnose a growing queue from metrics alone, and argue SQS against Kafka in a design review.

The approach

Every concept is built in the same order:

the problem → the intuition → how it actually works → what breaks in production → the real architecture

Two rules keep it honest:

Every number is verified. Quotas, limits and prices live in one place — _reference/aws-facts.md, checked against official AWS documentation and dated. Chapters link to it rather than restating values. Several widely-taught SQS "facts" are years out of date; misconceptions.md lists them.
Every "why" gets answered. Why deletion is explicit. Why duplicates happen. Why metrics say "Approximate". Why more consumers stops helping. Why FIFO needs a MessageGroupId. These are not optional asides — they are the parts that transfer to the next system you build.

Start here

If you are…	Start at	Then
New to messaging entirely	00 — Prerequisites	Read straight through
A backend engineer who knows queues	02 — Introduction to SQS	Then 04 → 05 → 08 → 09
Preparing for an interview	31 — Interview Preparation	Then modules 07, 08, 09, 13, 15, 25
About to own a queue in production	Production checklist	Then 08, 11, 17, 22, 26
Debugging right now	26 — Troubleshooting	Or the troubleshooting cheat sheet
Revising something you already learned	Cheat sheets	Pick by time available

Minimum path to being useful in production: 00 → 01 → 02 → 04 → 05 → 08 → 09 → 11 → 17

Difficulty key

🟢 Beginner · 🟡 Intermediate · 🔴 Advanced · ⚫ Production / System Design

Conventions

Code: Java 21+, AWS SDK for Java 2.x, currently supported Spring Boot. Never credentials in source — the credential chain, profiles, environment variables and IAM roles instead.
Diagrams: Mermaid, always followed by prose explaining them.
Numbers: sourced from _reference/aws-facts.md, never hard-coded in chapter prose.
Terms: defined once, in their owning module; linked everywhere else.
Depth: hard topics are marked 🟢 what is it / 🟡 how does it work / 🔴 what breaks and how do I design around it.

A note on accuracy

AWS changes quotas, prices and service behaviour. Everything in this course was verified against official AWS documentation on 2026-09-21, with sources recorded in _reference/aws-facts.md. That file carries its own re-verification checklist for the values most likely to drift.

If you find a discrepancy, the AWS documentation wins — and the fix belongs in that one file.

Design and reference

Course design Verified AWS facts Chapter template Glossary outline Common misconceptions Production readiness checklist Then 08, 11, 17, 22, 26 outline Cheat sheets Pick by time available

Part 1 — Foundations

00 — Prerequisites AWS account, credentials, CLI, Java 21 + SDK v2 Beginner 20 min 01 — Messaging Fundamentals The problem queues solve, derived rather than asserted Beginner 30 min 02 — Introduction to SQS What SQS is, and what it deliberately is not Beginner 25 min 03 — How SQS Actually Works Why every metric says "Approximate" — and what that implies Intermediate 30 min 04 — Sending and Receiving Messages The API surface; MessageId vs ReceiptHandle Beginner 35 min 05 — The Message Lifecycle The state machine every later module zooms into Intermediate 30 min

Part 2 — Core mechanics

06 — Standard Queues At-least-once, best-effort ordering, and how a duplicate is born Intermediate 30 min 07 — FIFO Queues What ordering actually guarantees, and what it costs Advanced 45 min 08 — Visibility Timeout The most consequential setting in SQS, and how to choose it Advanced 40 min 09 — Delivery Semantics and Idempotency Why exactly-once is impossible, and what to build instead Advanced 40 min 10 — Retries Backoff and jitter on a service that has neither Advanced 35 min 11 — Dead Letter Queues Quarantine, alarm, runbook — and the retention trap Advanced 35 min

Part 3 — Efficiency and scale

12 — Long Polling One setting that cuts cost ~10× and latency at the same time outline Intermediate 25 min 13 — Batch Processing API call vs message vs batch vs consumer vs worker, untangled outline Intermediate 40 min 14 — Scaling Consumers When adding consumers helps — and when it makes things worse outline Advanced 40 min 15 — Performance and Throughput Little's Law, drain time, and the arithmetic for a design review outline Advanced 45 min

Part 4 — Operating it

16 — Security Identity vs resource policies, KMS, cross-account outline Advanced 40 min 17 — Monitoring and Observability Every metric, and why queue depth alone tells you nothing outline Advanced 45 min 18 — Cost Why billing is per request in 64 KB chunks, and what that changes outline Intermediate 25 min

Part 5 — Integrations

19 — SQS and AWS Lambda Who actually polls, how it scales, and the whole-batch-retry trap outline Advanced 45 min 20 — SNS + SQS Fanout Adding a consumer without touching the producer outline Advanced 35 min 21 — Event Sources: S3, EventBridge, API Gateway Producers you did not write outline Advanced 35 min 22 — Self-Managed Consumers on ECS, EKS and EC2 The poll loop, and the graceful shutdown that deploys depend on outline Advanced 40 min 23 — SQS with Java and Spring Boot @SqsListener — and a hard line between Spring's behaviour and SQS's outline Advanced 45 min

Part 6 — Synthesis

24 — Distributed Systems Concepts What is universal vs what is AWS-specific outline Expert 40 min 25 — Production Architecture and System Design A 20-question methodology and a reference architecture outline Expert 50 min 26 — Troubleshooting 18 scenarios: symptom → cause → verify → fix → prevent outline Expert 50 min 27 — SQS vs Kafka vs RabbitMQ An architectural comparison, not a feature table outline Expert 45 min 28 — Corporate Case Studies Six complete systems, each with a real production failure outline Expert 60 min

Part 7 — Practice

29 — Hands-on Labs 17 parts 17 labs; watch the mechanisms happen instead of reading about them outline Intermediate 30 — Capstone Project 12 parts One production-grade order platform exercising everything outline Expert 31 — Interview Preparation Question bank, answer frameworks, 12 design scenarios outline Expert

Revision pack

Revision pack index 9 parts outline