ReferenceGlossary
OLTP and OLAP
Online transaction processing (OLTP) runs many small, fast reads and writes on current rows, such as orders and logins. Online analytical processing (OLAP) runs few large scans and aggregations over history, such as reports. OLTP databases store each row together; OLAP engines usually store each column together, so a scan reads only the columns it needs.
Data
In Quorum#
Report requests send q-scan queries. On the Database a scan costs a lot of CPU and evicts its buffer cache, so other reads go to disk and writes wait; the Report kills the database Incident scans 10 times as much. A Data Warehouse (Data warehousing research) takes the scans off the primary: 1.5 s each, no capacity limit, and $0.02 per scan times the Era's cost index.