Serverless in production, an experience report

In this talk Yan Cui shares his experience of migrating an existing monolithic architecture for a social network to AWS Lambda, and how it empowered a small team to deliver features quickly and how they address operational concerns such as CI/CD, logging, monitoring and config management.

This sessions draws on my experience leading a team that transformed our entire architecture in under 6 months, where we also managed to cut cost by over 90%, and increased production deployments from 4-6 per month to 80-100 per month. We did that whilst coming up with solutions to those common operational concerns to ensure we not only run a serverless architecture in production, but we do so responsibly.

The audience should have basic understanding of how AWS Lambda works, and are aware of some popular AWS services like DynamoDB, Kinesis, S3. I’m the best person to speak on the subject as I lead the team on this migration and were responsible for large parts of the actual implementation, and as the architect I was also responsible for our entire architecture (and the first one to be woken up at night if things go wrong!). I have written extensively on the operational aspect of working with AWS Lambda, and my work has been regularly referenced by the AWS Lambda team including talks at re:invent, as well as the new Serverless Well Architected whitepaper.