From 0 to 1.000.000: How to build a serverless raffle service for hyperscale

How do you build a serverless app for one million users? Discover why synchronous patterns fail and how an asynchronous architecture can easily handle massive traffic bursts.

#1about 3 minutes

Understanding the challenge of hyperscale serverless applications

Hyperscale applications must handle sudden traffic bursts of millions of users and hundreds of thousands of requests per second, a common scenario for raffles or voting systems.

#2about 7 minutes

Identifying bottlenecks in a standard serverless architecture

A typical synchronous architecture using API Gateway, Lambda, and DynamoDB fails at hyperscale due to Lambda's burst concurrency limits and the inefficiency of single-request processing.

#3about 4 minutes

Building a decoupled architecture with SQS for scalability

Decoupling the architecture by using an SQS queue between API Gateway and Lambda allows the system to buffer incoming requests and process them asynchronously in batches.

#4about 10 minutes

Demonstrating the architecture with a live load test

A live demo using the Artillery load testing framework and audience participation successfully subjects the asynchronous architecture to over 100,000 requests per second.

#5about 2 minutes

Summarizing key takeaways for building hyperscale systems

Building for hyperscale requires understanding service limits, leveraging asynchronous patterns, writing efficient code, pre-warming databases, and performing rigorous load testing.

#6about 4 minutes

Answering audience questions on serverless architecture

The speakers address when to use queues versus synchronous calls, how to handle requests needing immediate validation, and best practices for throttling with exponential backoff.