Improving the Performance of Data Warehousing with Column Databases
Anh V. Nguyen, Stephanos Mavromoustakos · 2024
Databases have traditionally been used for different kinds of data, however, all have the same basic access patterns, known as online transaction processing (OLTP), that are similar to processing business transactions, such as making a sale, paying an employee's salary, or placing an order - the application most of the times looks up for only a small number of records then read, update, insert, or delete it based on user's input. However, as technologies grow, databases also begin to be increasingly used for another purpose - data analytics - whose query patterns are completely different from the former and demand a different approach. In this paper, we analyze the reason why traditional row-oriented databases, while can be used for both purposes, struggle the latter in terms of speed, and how columnar databases come out as a great solution.