The first pattern published on www.daxpatterns.com is the Cumulative Total. Another common name of this calculation is Running Total, but the reason why we named the pattern in this way is that we identify all those scenarios in which you want to obtain a value, at a certain date, that corresponds to the result of a number of transactions executed in the past. Usually, this scenario is implemented using snapshot tables in the classical Kimball modeling. With a columnar table, you can afford avoiding the snapshot table even with a relatively large dataset.
You might want to implement the Cumulative Total pattern to reduce the volume of data stored in memory, transforming snapshot tables into dynamic calculation in DAX. The examples shown in the article represent an implementation of the Inventory Valuation at any point in time. Remember, I am not saying snapshot tables can be avoided in Tabular, but you have to consider the alternative, especially when the size of snapshot table is an order of magnitude (or more) larger than the original transactions table. Do some benchmark and find the better solution for your model!
Originally appeared on: http://sqlblog.com/blogs/marco_russo/archive/2014/02/20/the-cumulative-total-dax-pattern.aspx