Ravindra BagaleCourses & study guides मराठी Track your progress

Project: Build an Async Order API with SQS, Lambda and DynamoDB (Source Code)

By , trainer LinkedIn Last updated:

Flipkart flash-sale style: accept the order in milliseconds, process a moment later. Intake puts work on SQS; a worker Lambda writes DynamoDB.

What you'll learn in this chapter

  • Why a queue in front of workers
  • SQS + worker + table
  • Return 202 Accepted
  • Cleanup

1. Everyday example

Why do we need this?

If hundreds hit Place order together, a slow write path times out the API. SQS absorbs the spike; workers drain it.

What is it?

HTTP API → intake Lambda → SendMessage → SQS → worker Lambda → DynamoDB PutItem. Client gets 202.

Async pathClientPOST /ordersIntakeSendMessageSQSbufferWorkertriggerDynamoDBrow
API stays fast if worker is slow.

Download: async-order-api.zip

2. Lab

How to do it (step by step)

A. Queue + table

  1. SQS Standard queue yourname-orders. Copy URL.
  2. DynamoDB yourname-orders, PK orderId, on-demand.

B. Worker

  1. Role: logs + SQS consume + PutItem.
  2. Function worker.py, env TABLE_NAME. Add SQS trigger batch size 5.

C. Intake API

  1. Function intake.py, env QUEUE_URL, IAM sqs:SendMessage.
  2. HTTP API POST /orders. CORS *.
curl -s -X POST "$API_BASE/orders" -H 'content-type: application/json' \
  -d '{"sku":"TEE-RED","qty":2}'
# wait ~2s — DynamoDB item status accepted

Cleanup

  1. Remove SQS trigger, both Lambdas, API, purge+delete queue, table, roles.

3. Interview

Q. Why 202 not 201?

Accepted for async processing, not created finished.

Q. Standard vs FIFO?

FIFO for strict order + dedupe; Standard fine here.

Q. Worker failure?

Visibility timeout retry; DLQ after maxReceiveCount.

Q. Native API GW to SQS?

Yes — fewer parts; lab uses intake for clear code.

Q. Cleanup?

Queue, Lambdas, API, table.

Try at home

  1. Attach a DLQ after 3 receives.
  2. GET /orders/{id}.

In short

  1. Async = intake → SQS → worker → DynamoDB.
  2. 202 keeps API fast.
  3. Triggers + IAM matter.
  4. Clean queues too.

Practice Tasks

About this scenario

Dubai Gold Souk is a jewellery business in Dubai. Fatima Al Mansoori looks after its IT and cloud work. They count money in AED and run their AWS resources in the me-central-1 Region. In the tasks below you do real jobs for them using what this lesson teaches. All names, account IDs, IP addresses and numbers are made up for practice.

0 / 12

Easy List functions

Why: See which functions exist in this Region.

What to do: List the Lambda functions in me-central-1.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Medium Invoke a function

Why: Invoke runs the function now and saves its answer in a file.

What to do: Invoke dubaigold-orders-fn and save the response to out.json.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Medium Trust policy for a Lambda role

Why: The trust policy says WHO may take the role.

What to do: Write a trust policy that lets lambda.amazonaws.com assume the role.

How:
  1. Start with { "Version": "2012-10-17", "Statement": [ ... ] }.
  2. In each statement write Effect, Action and Resource.
  3. Give only what the task needs (least privilege).
  4. Use double quotes and commas correctly. Click Check answer.

Medium Put an item

Why: DynamoDB JSON shows the type of each value: S for string, N for number.

What to do: Put an item into dubaigold-orders with orderId "o-101" (string) and amount 2500 (number).

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Medium Change timeout and memory

Why: More memory also gives more CPU; timeout stops runaway runs.

What to do: Set dubaigold-orders-fn to timeout 30 seconds and memory 512 MB.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Medium Let a function write logs

Why: Without these actions, Lambda logs never appear in CloudWatch.

What to do: Allow logs:CreateLogGroup, logs:CreateLogStream, logs:PutLogEvents on arn:aws:logs:*:*:* only.

How:
  1. Start with { "Version": "2012-10-17", "Statement": [ ... ] }.
  2. In each statement write Effect, Action and Resource.
  3. Give only what the task needs (least privilege).
  4. Use double quotes and commas correctly. Click Check answer.

Hard Connect an SQS queue to Lambda

Why: An event source mapping makes Lambda poll the queue and process messages in batches.

What to do: Create an event source mapping from queue arn:aws:sqs:me-central-1:112884599412:dubaigold-orders to dubaigold-orders-fn with batch size 10.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Hard Create a Lambda function

Why: Lambda needs the code zip, a runtime, a handler name and an execution role.

What to do: Create function dubaigold-orders-fn: runtime python3.12, handler app.handler, role arn:aws:iam::112884599412:role/lambda-basic, code in function.zip.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Hard Set environment variables

Why: Keep settings like table names out of the code.

What to do: Set the environment variable TABLE=dubaigold-orders on dubaigold-orders-fn.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Hard Invoke with a JSON payload

Why: In CLI v2, raw JSON needs --cli-binary-format raw-in-base64-out.

What to do: Invoke dubaigold-orders-fn with payload {"orderId":"o-101"} (raw JSON) and save the reply in out.json.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

Hard Lambda GB-seconds

Why: Lambda bills memory × time.

What to do: 1,000,000 invocations, 512 MB memory, 200 ms each. How many GB-seconds?

How:
  1. Write down the numbers the task gives you.
  2. Work it out step by step on paper.
  3. Type only the final answer in the box.
  4. Click Check answer.

Hard Let S3 call your function

Why: A resource-based permission is needed before S3 can invoke the function.

What to do: Add permission statement s3invoke to dubaigold-orders-fn so s3.amazonaws.com can call lambda:InvokeFunction from bucket arn:aws:s3:::dubaigold-uploads.

How:
  1. Read the task and decide which AWS service it is about.
  2. Write aws, then the service name, then the operation (the action word).
  3. Add each needed flag (--name) followed by its value from the task.
  4. Flags can be in any order. Click Check answer.

See all practice tasks in the Task Hub →

Revision, comparison and interview cards

Revise this lesson in a few minutes. Save the cards on your phone or download them as PNG files. The QR code on each card opens this lesson.

Revision card: Why an async order API?. If checkout waits for inventory, fraud checks, and email, the shopper stares at a spinner and retries. Async intake: API accepts the order fast (HTTP 202), puts a message on SQS, and a worker Lambda writes DynamoDB later. Build SQS queue + DynamoDB PK orderId + intake Lambda (SendMessage) + worker Lambda (SQS trigger) + POST /orders. curl POST {"sku":"TEE-RED","qty":2} returns 202 in ~50 ms; ~2 s later DynamoDB shows status accepted.
Revision card: Why an async order API? · Download PNG · Open card page
Revision card: SQS as a buffer. A sudden flash sale can drown a single Lambda concurrency limit. Work must wait safely. Amazon SQS is a managed queue. Producers SendMessage; consumers receive later. Standard queue is fine for this lab. Create queue yourname-orders, grant intake sqs:SendMessage, grant worker consume + PutItem, add event source mapping batch size 5. If the worker is slow, messages pile in ApproximateNumberOfMessages; the API still returns 202.
Revision card: SQS as a buffer · Download PNG · Open card page
Revision card: HTTP 202 vs 201. Status codes tell clients what “success” means. Wrong codes confuse frontend retries. 201 Created = resource finished. 202 Accepted = request accepted for processing that is not done yet. Intake Lambda returns statusCode 202 and a body with orderId; worker creates the DynamoDB row asynchronously. Mobile app shows “Order received” immediately from 202, then polls GET /orders/{id} when you add it later.
Revision card: HTTP 202 vs 201 · Download PNG · Open card page
Comparison: Sync API to DynamoDB vs Async API via SQS. Sync API to DynamoDB: Caller waits until the row is written. Async API via SQS: Caller gets 202; worker writes later.
Comparison: Sync API to DynamoDB vs Async API via SQS · Download PNG · Open card page
Interview card: When do you return 202 Accepted?. When the server accepted work that will finish asynchronously. 201 would claim the order resource is fully created already. Intake SendMessage to SQS → 202 + orderId; worker writes DynamoDB afterward.
Interview card: When do you return 202 Accepted? · Download PNG · Open card page
Interview card: What happens when the worker keeps failing?. After visibility timeout the message reappears. After maxReceiveCount it can go to a dead-letter queue (DLQ). DLQ protects the main queue from poison messages and keeps a copy for debugging. Attach redrive policy maxReceiveCount=3 to an SQS DLQ on yourname-orders.
Interview card: What happens when the worker keeps failing? · Download PNG · Open card page
Interview card: Standard SQS or FIFO for this lab?. Standard is enough when strict ordering and deduplication are not required. FIFO adds ordering per message group and exactly-once processing features at lower throughput. Checkout intake lab uses Standard; use FIFO when order-of-events per customer must be strict.
Interview card: Standard SQS or FIFO for this lab? · Download PNG · Open card page

More cards: Ravi Sir's Cards.