Understanding the data analytics system
Data originates from transactional systems such as e-commerce systems or social networking sites. These systems process live data actively and thus are called Online Transaction Processing (OLTP) systems. They typically store their data in operational databases, aka transactional databases, that are built for fast transaction processing and integrity. These databases are optimized for fast data storage and retrieval, typically focusing on shallow datasets (datasets without complex relationships), most-recent events, and deduplicated data. When the transactions complete, the data will be moved out of the operational databases to make room for more recent data.
As important as they are, transactional databases are not designed for analytical purposes. They are designed to handle datasets for transactions, usually with limited size, duration, and context. Analytics-oriented systems usually answer general business questions, which involve longer...
