Here's a simple scenario that leads to the OOM:

./bin/juttle -e "read stochastic -source 'logs' -from :2014-01-01: -nhosts 100 -lpm 1000 | write file -file 'test.json'"

From a quick look at the code its waiting to accumulate all of the data points before writing them out to the underlying file instead of writing out each point or batches of points in order to avoid running out of memory.