Understanding AWS Lambda: Invocations, Events and How It Really Works

AWS Lambda is often described as “run code without servers” — but the real power (and confusion) lies in how Lambda is invoked.

There are three core invocation models every developer should understand:

• Synchronous invocations — request/response style, commonly used with API Gateway
• Asynchronous invocations — event-driven, retried automatically, great for background jobs
• Event source mappings — Lambda polls services like SQS, DynamoDB Streams, and Kinesis

Each model has very different behavior when it comes to retries, error handling, concurrency, and cost.

Once you understand:

• how Lambda scales,
• when cold starts happen,
• and how retries actually work,

you stop treating Lambda as “magic” and start designing simpler, more resilient systems.

👉 I wrote a detailed breakdown here (with real-world context, not just docs):
Read the full article: here