The Mission of the Business Intelligence Performance Index (BIPI)

Users of Business Intelligence (BI) tools and analytical databases, such as data warehouses and data marts, are not interested in reviewing, trending, graphing or manipulating large amounts of consistent data - whether that consistency is represented by a repetitive list of a constant value or a consistently random set of data.

In the case of constancy, there is little value to be gained from a BI tool over a simple, naïve query, and little business or process insight to be gained from what everyone already knows anyway (“it’s all the same”).  Likewise, a set of consistently random data is, by definition, just “noise” and provides even less insight.

Business Intelligence users and analysts are looking for trends, patterns and anomalies in their data.

•    North American grocery stores do sell more eggnog in December.  
•    Some retail sales promotions do (and some don't) drive increased store traffic.  
•    Home heating oil consumption does increase during winter in colder climates.  
•    More television viewers do watch sports specials than the daily newscasts.  
•    Marketing campaigns can drive increased traffic to a particular web site.

These are the patterns in real business, system and process data that are of interest to real users and therefore the environment upon which BI software should be benchmarked.  This is in stark contrast to other benchmarks that purposely use randomized and/or randomly generated data that is more fitting for a physical test of hardware systems into which the data is loaded than the BI tools that will be used.  

We all know that data doesn’t get created, captured and stored in smooth, consistently random fashion, so the BIPI doesn’t propose a test for those tools where the first and most fundamental choice for the test environment – the data – is not aligned with users’ realities.  And as the name suggests, this is a performance-oriented index, not strictly, or even mostly, a functionally-oriented set of tests but must represent real-word BI functional use cases.

Simply put, the data matters.  

Stay tuned for The BIPI - Part II, The Data