On-line analytical processing is a commonly heard term in business intelligence. But exactly what is OLAP?
When you are evaluating business intelligence tools, this term will often come up. Some of the confusion may lie in that vendors will often have different concepts of what it means.
The term first came about in the early 1990s by a researcher named Edgar Codd. He had worked on what would later be called the Essbase software package, which later became part of Hyperion and then currently Oracle’s BI offering. While the 12 rules offer a good overview of the characteristics, it was quickly part of controversy since it was directed toward the Essbase product rather than an overall industry standard. Still, the rules are interesting to know. There is also kind of an unwritten rule zero which states that the data should be in an RDBMS format. Interesting, because many OLAP tools today use the cube format for data structure. You can see how the definition has been twisted around a bit over the years, so it is tough to decide with any clear outcome whether or not software is “compliant”. It does give a little bit of background into the technology and theory though. I’ve attempted to summarize them here.
1) Information can only be provided as columns in the relational tables.
2) All data should be retrievable, that is, it should use primary keys
3) Null values should be allowed
4) A data catalog must be available
5) It must allow a RDBMS language for queries
6) Views should be able to be refreshed
7) Data deletion, insert, update must be available.
8) Logical layer must be separate from the physical layer.
9) Changing the logical layer should not affect the application layer
10) Data should be separate from other programs
11) Data integrity should be separate from other programs
12) System interface should not be used to subvert the system
Besides this, the FASMI test has also gained popularity because of its simplicity.
Copyright 2011-2019 BusinessIntelligenceBase.com