Columnar storage is a commonly used storage technique. Often, it implies high performance and has basically become a standard configuration for today’s analytical databases.
The basic principle of columnar storage is reducing the amount of data retrieved from the hard disk. A data table can have a lot of columns, but the computation may use only a very small number of them. With columnar storage, useless columns do not need to be retrieved, while with row-wise storage, all columns need to be scanned. When the retrieved columns only take up a very small part of the total, columnar storage has a big advantage in terms of IO time, and computation seems to get much faster.