Data warehouses, data lakes, data marts and cloud object storage (if you do not consider this to be another form of data lake) all exist to support analytics and query processing. Many companies have all of these and frequently several of some of them. Each can individually support relevant analytic functions. However, what happens when you want to run queries that span these sources? You essentially have two solutions: you either query the data where it is and then join the results or you move the data so that it is all in a common environment (a hybrid approach is also possible). In this paper we will examine the reasons why you might choose one approach or the other.