What is this OLAP thing?
Almost a year ago, I spent an entire week on the ultimate Business Intelligence course. I have taken and passed the corresponding brainbench tests (Data warehousing and OLAP). Still I must admit that I am not confident about OLAP (especially MDX). Having lots of stuff boiling in my sub-conscious mind, I will get there eventually (maybe I am a bit slow, he he). And my guess is that any BI specialist confident about their understanding of leveraging OLAP technology, has stopped developing.
Wikipedia (initially) states
“OLAP is an acronym for online analytical processing. It is an approach to quickly provide the answer to complex analytical queries.”
google define gives eg.
On-line retrieval and analysis of data to reveal business trends and statistics not directly visible in the data directly retrieved from a data warehouse. Also known as multidimensional analysis
On-line Analytical Processing. Information analysis that pass the FASMI test: Fast Analysis of Multidimensional Information.
FASMI: Fast Analysis of Shared Multidimensional Information. The summary description of OLAP applications used in The OLAP Report, and now very widely referenced elsewhere.
http://www.olapreport.com/fasmi.htm: We think that the FASMI test is a reasonable and understandable definition of the goals OLAP is meant to achieve.
So, here some of my own definitions (it depends on what angle you want to see it, of course):
OLAP (personal): An interesting possibility I would like to be more involved with.
OLAP (informative): A tool for converting data into actable information.
OLAP (technical): Information analysis that is fast, statistical, multi-dimentional, ad hoc, secure and built on a large amount of data.
OLAP (for the dba): All the groupings and aggregates you can do on a dataset precalculated.
OLAP (measurable): Analysis leveraging a tool that passes the FASMI test.
Tools must give most answers in less than five secounds
It should be possible to add value to the data, converting it to information, being data mining, statistical formulas, score-carding, exception alerting or other special applications
Integrate into the security model and following any complaience requirements
This is not hard to understand: it must take into account more than two measures at a time, eg. the three or four: "what is it", "where is it", "when is it" and possibly "is it".
Much information. It must be able to gather, manage and use much information.
Thanks goes to two undisputed geniuses
Reed Jacobsen, genius, Hitachi
Kamal Hathi, Lead Program Manager (=genius) for SQLIS (DTS/ETL) (SQL Server 2005) and (Lead, I think) Program Manager for OLAP (SQL Server 2006+), Microsoft