Anticlue

Comparison of Transaction and Decision Systems

For the customer who aspires to be a technical guru, sometimes the basics are not so basic. In a recent inherited project, I encountered an architectural design flaw in the system when it was about to be 70% completed. The issue at hand was that for decision reporting was being run against a transactional system. The reports were taking a long time to display to the customer, and the creation load was also causing issues with the transactional system. This issue was discovered in integrated testing the new system. So a clarification of the difference between a transactional and decision systems was needed, and a reworking of several key components of the project.

In a transactional system, the architectural structure is focused on data entry response time. The data is current, historical data should be archived. The data structure is at the detailed and granular level. The data organization is to perform quick queries, updates, and deletes upon a single record. Information is updated continuously through out the day.

In a Decision System, the architectural structure is focused on reporting response time. The data is historical information. The data structure is at a summarized, aggregate level. The data organization is to perform quick queries on numerous records at a time. Information is read-only, updated through a nightly extraction, transformation, and load from the transactional systems.