This was a talk I gave at Erlang User Conference 2015 in Stockholm.
This presentation describes the experience gained while rewriting one of the first systems developed in Erlang at bet365. It offers insights into the architecture of a busy message publishing system and how code quality improved as we accumulated more Erlang expertise.
Background
At bet365, we had one of our early Erlang systems that had grown organically over time. As our team's understanding of Erlang deepened and best practices evolved, we recognised the opportunity to rewrite the system with the benefit of hindsight.
The system in question was a high-throughput message publishing system - the kind of real-time, concurrent, distributed system that Erlang excels at.
Key Lessons
The talk covers:
- How our coding practices evolved as we gained more experience with Erlang/OTP
- Architectural decisions for handling high message volumes
- Patterns that worked well and anti-patterns we learned to avoid
- The practical challenges of rewriting a system while maintaining service
Watch the Talk
The video and slides are available on the Erlang User Conference 2015 page.