Editorials

Best Practices for Business Intelligence – the Complexity Curve

Featured Article(s)
Book Review: Practical SQL Handbook: Using SQL Variants
A review of the book – it’s targeted at the beginner DBA, and part of a series of titles for this market.

Best Practices for Business Intelligence "Complexity Curve"
Peter wrote in with thoughts from his perspective – and specifically asking (paraphrasing) what the benefit is for third-party tools if proper planning is done – are native tools enough? I must say, I’ve seen some very compelling solutions using third-party tools. The presentation of information, the visuals and manipulation ability – they all add some valuable utility to the overall process. That said, it’s also true, I believe, that gathering and storing information is quite different from the mentality of accessing and using that information.

From Peter: "I am a finance guy at an airline, yet my BI experience started long time ago. I was working for Oracle as an intern, and did some basic ETL with csv files and Excel VBA macros. You can do a lot, but as you stated, there is a sharp drop-off in terms of flexibility when you go beyond the odd one-off/stand-alone report.

That said : it did what it had to do : provide management with some graphs to highlight the issue. Based on that, we started experimenting with solutions, and ultimately fixed the issue. There is nothing more to it, we didn’t need cubes, no dynamic querying, just some copy-pastes from an Excel graph in a PowerPoint presentation.

Since then, my experience has been in environments that have operational applications running on Oracle DBs. The Microsoft BI is used on top, mostly because the ease of entry of the SQL server all-in-one license. That said, it is quite a powerful toolset, and it allows business people (like me) to get quite deep into the inner workings of the DWH. I use SSIS extensively, for ETL and then the pre-processing of data into fact- and dimensional tables. Not so much for SSAS purposes, but also for high performance from the SSRS side. The unified DWH allows me to quickly provide new reports, and to trace back findings that are questioned without problem. Also, it allows me to generate consistent reports (reporting the same numbers from different angles), and to benefit from the modular set-up : if we find a gap in some number, I need to fix one of the ETL trails all the way until the relevant fact/dim table. Once that is done, all the other reports will also benefit from the improvement/fix.

SSRS is my favorite tool for passing information to the end-user, as they need only minimal understanding of what is behind the numbers (sideline : operational DBs are always structured conceptually different from the perspective business people have; they can’t relate to the data behind the scenes; SSAS oftentimes will open the door to peek behind the scenes, which may confuse end-users). SSRS covers 95% of management needs, and I feel any further effort is better spent in widening the scope of the DWH, than in improving user interface of graphics. In my opinion, [third-party] tools are oversold based on sales-pitches to senior management who fail to understand the split between the process of gathering the data and presenting it. As if somehow [third-party tools] as a presentation tool would generate better results regardless of the resources put into the preparatory work. And surely, for that price, it should…

So, SSIS and SSRS, a tiny bit of SSAS, and you would be hard pressed to find a valid business need for anything more complex."

Featured White Paper(s)
Improving Data Protection and Storage Reliability for Critical Databases
Discover how a healthcare management company met the challenge of improving data protection and storage reliability for criti… (read more)