A decade later I have learned quite a lot implementing event sourced systems, and I would implement a few things a bit differently nowadays. In particular, I would recommend use a separate table for event streams and an N:M mapping between events and stream, such that each event can be part of multiple streams. This is not a must, though, as I have successfully implemented several systems that are in production right now that use a similar table layout as this solution.
I'm no longer developing this event store and therefore merely keep it as a reference implementation for other people to learn about event sourcing and investigate.
[Complete] JPA 2.0 (any supported database)
[Planned] JPA 2.0 w/ Master (writes) / Slave (reads) support
[Planned] MongoDB
[Planned] File based storage
[Planned] In-memory
[Complete] JSON w/ Google Gson
[Planned] Memcached deserialization cache
[Planned] Encryption layer