[bonobo3000]

Yeah i have seen Mesos used for managing big data frameworks like Spark. Any large distributed computing framework has to deal with some of the same problems - managing resources on worker nodes, sending tasks, getting results, scheduling etc. and Mesos/YARN provide a common base here, better than each framework duplicating the same code and solving the same problems.

[StavrosK]

That makes sense, thanks. The workers are pretty independent from each other, too, so this setup would be more fitting for that.