AWS Lambda Triggers: S3, EventBridge, SQS and SNS Explained

By Ravindra Bagale, trainer · LinkedIn · Last updated: 23 October 2026

1. A story: photos that resize themselves

Meera's campus app stores student photos in S3. Full-size uploads are 5 MB. The student list needs 200 KB thumbnails. She does not want a server watching a folder. She wants: photo lands in S3 → thumbnail appears.

That is an S3 trigger on ObjectCreated, pointing at a Lambda that reads the object, resizes it, and writes to a thumbs/ prefix.

2. Trigger map

3. Cold starts, timeouts and retries

Why do we need this?

The first request after idle can be slower. Timeouts cut runaway work. Retries can double-charge or double-write if your code is not safe.

What is it?

• Cold start: AWS creates a new execution environment, pulls your package, starts the runtime. Then your handler runs. Later calls on the same environment are warm and faster.
• Bigger packages and Java/.NET often cold-start slower than Python/Node. Provisioned concurrency keeps environments warm for a price.
• Timeout: function setting, max 15 minutes. Set it above your p99 duration or real work dies mid-flight.
• Retries: async invokes (S3, SNS, EventBridge) retry twice by default. SQS uses the queue redrive policy. Make handlers idempotent (safe to run twice).

4. Layers and environment variables

What is it?

A layer is a zip of libraries you attach to many functions (up to 5). Example: a Pillow layer shared by three image functions. Keep the function zip small.

Environment variables store config (bucket names, log level). Total size max 4 KB. Do not put passwords here — use Secrets Manager or SSM.

5. Lab A: S3 → Lambda (notify only)

How to do it (step by step)

Full Pillow resize needs a layer; this lab proves the trigger with a log line. Region ap-south-1.

1. Create bucket yourname-lam-pics, Block Public Access on.
2. Create Lambda s3-notify-lab, Python 3.12, basic execution role.
3. Code:

def lambda_handler(event, context):
    for r in event.get("Records", []):
        b = r["s3"]["bucket"]["name"]
        k = r["s3"]["object"]["key"]
        print(f"uploaded s3://{b}/{k}")
    return {"ok": True}

4. Deploy. Configuration → Permissions → role → Add permissions → Attach policies → AmazonS3ReadOnlyAccess (lab only). Why: later labs that download need s3:GetObject.
5. Lambda → Add trigger → S3 → bucket yourname-lam-pics → Event type All object create events → Suffix .jpg → Add. Confirm the recursive warning (do not write back to the same prefix you listen on).
6. Upload: aws s3 cp cat.jpg s3://yourname-lam-pics/raw/cat.jpg.
7. CloudWatch logs for s3-notify-lab should show uploaded s3://yourname-lam-pics/raw/cat.jpg within a minute.

6. Lab B: EventBridge schedule → Lambda (EC2 echo)

In the CloudWatch EC2 alarm chapter you stopped idle instances with alarms. Schedules are cleaner for fixed office hours. This lab only prints; add StartInstances/StopInstances when ready.

1. Create Lambda ec2-schedule-echo that prints event and returns ok.
2. EventBridge → Schedules → Create schedule. Name office-open-echo. Recurring — Cron: 0 9 * * ? *. Time zone Asia/Calcutta. Why: office 9 AM India.
3. Target: AWS Lambda Invoke → pick ec2-schedule-echo → Create.
4. For a quick test use rate(5 minutes), watch logs, then switch back or delete.

7. SQS and SNS in one minute

• SQS trigger: Lambda polls the queue for you (event source mapping). Good when producers are bursty and you want retry + DLQ.
• SNS trigger: publish once, many subscribers (Lambda, email, HTTPS). Good for alarms and fan-out.
• Permission: the trigger adds a resource-based policy on the function so that service may invoke it.

Cleanup

1. Delete schedules, remove S3 trigger, delete functions, empty and delete the bucket.
2. Detach extra IAM policies; delete unused roles.

8. Interview questions

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.

More cards: Ravi Sir's Cards.