If you do end up going with the Kafka route, there is a native Haskell client: https://github.com/tylerholien/milena Arnaud Bailly writes:
Hello,
I am implementing an application using event sourcing as primary storage for data, which implies I need a way to durably and reliably store streams of events on stable storage. I also need to be able to have an event distribution system on top of that persistent storage so that components can subscribe to stored events.
So far I have implemented a simple store, e.g. a flat file, which reuses the format of Apache Kafka (just in case...). Not very robust nor sophisticated but can work for moderate loads. Now I am looking for the event distribution part in the hope of being able to reuse some distributed event bus system that might exist somewhere and not having to roll my own.
I have had a look couple of months ago at Vaultaire, Marquise and friends, but I am not sure they are really suited to my use case: They seem to be geared toward very high workload and throughput, like log or huge data streams analysis.
Thanks for any pointer you might share,
-- Kyle Marek-Spartz