Columnar Data Model
Unlike traditional relational databases, which are optimized for storing rows of data and focused on transactional applications, the columnar data model is ideal for solutions that perform large-scale operations on the database, such as aggregations and calculations with a high volume of data reads.
It is primarily applied to analytical queries as it reduces disk I/O requirements. Direct reading of columns, data compression, and projections make the response time in massive calculation and aggregation operations shorter in this model.
For example, a select avg in a row-oriented database will load all fields to perform the operation and return the average.
In the columnar database, only the column to be processed (with the avg operation) will be evaluated, thus consuming fewer resources.
The most suitable type of system for using columnar storage is one that demands reads and operations on large volumes of data.
