Ask HN: What sort of data do distributed systems handle well?

That's pretty much the question. Imagine you had access to whatever massive data set you want, live or archival, what would would fit with something like improbable.io?