Hello,
I have been "playing" with SQL Server 2014 clustered columnstore indexes and found that they can be placed on a partition scheme. I am having hard time understanding how this might physically work since the data is obviously stored in columns. With regular row-oriented storage the underlying mechanism is more or less clear. We are basically horizontally partitioning data and subsequently can benefit from "partition elimination". With columnstore the data is stored in columns. Does this mean that the columns are further horizontally segmented? Let's say I have a table with 20 columns. One of the columns is used as partitioning key. How will this column, as well as all of the other columns be stored? Can I still benefit from "partition elimination"?
Thank you!