The modern software industry is fairly resistant to the idea of hand-rolling solutions to common problems, but in this talk I argue that you’re better off building your own event store in the first instance rather than relying on commercial products or large open source frameworks to do it for you. Drawing from my experience implementing an event sourced solution after being a long time skeptic, I’ll talk about the various tradeoffs involved in building an event store, and show how your technology stack and application requirements will affect the design. Some of the considerations include: * Event payload storage & versioning * Different snapshot approaches * Strongly consistent vs eventually consistent queries * Event streaming/publishing mechanisms * Reactor/projector implementations and efficiently rebuilding read models Attendees don’t need any existing event sourcing experience, but some general database knowledge is assumed. The talk is suitable for anyone who has thought about adopting event sourcing in their systems, or just interested in how others have gone.