Introduction

When your data no longer can reside on a single machine, you're forced to apply distributed computing. tools like maproduce and spark are certainly available and powerful for these situations. But they don't give us an obvious, out of box way of dealing with time series. A distributed time series is just e time serieus that ise outer lon necambulac, billions of observations. So we have to make a modela really works on such distributiv system.