Data Warehouse Failure Statistics

Is there any recent study / statistics / surveys on data warehouse project pitfalls and risks ?
Your help is truely appreciated.


Best Regards,

I don't have any current references, but Gartner and TDWI have published papers on this topic in recent years. I think some of the other industry analysts have as well. In general, the main problems always seem to be soft in nature: lack of business requirements, poor business sponsorship, lack of user involvement, bad/no methodology, etc. I'm sure this comes as a shock to anyone who's read anything we've ever written. ;-)
--Warren

warrent wrote:In general, the main problems always seem to be soft in nature: lack of business requirements, poor business sponsorship, lack of user involvement, bad/no methodology, etc.

Three Ps - Personalities, politics and Power struggles. I have just been victimised by these forces in my last two BI projects when trying to apply best practices. You may not believe it, using Kimball's mini-dimension concept to address multimillion monster dimension could lose your job only because some DBA did not like the concept, even though it has delivered a perfect solution for the business.

Most BI/DW projects failed not because of technical reasons, but because of political reasons. Huge amount of money can be wasted for the political correctness while little effort can be put to address core issues.

I think it always helps to bring someone from the Mainframe world to head up a DW, which has been the case in the last 2 places I worked. And then rely on guys from the transaction side to dictate design.

Why should people from Mainframe world and transaction side dictate the design to help. I thought their comfort zone is in legacy system not BI/DW world. I believe the trust should be on BI/DW professionals to deliver viable BI solutions.

Whenever a BI/DW design is dictated by knowledge only applicable in old word, the architecture becomes the recipe for failure. Mainframe knowledge is good skill, but it's very old, older than Client/Server technology which is the foundation of RDBMS. Similarly, relational thinking from transaction side is very useful in operational system but needs to be replaced by dimensional thinking when it comes to the BI/DW world.

I am not saying people from old world cannot become front runners in the new world, and actually most, if not all, best BI/DW experts have evolved themselves from the old world. All I am saying is people need to realise the paradigm shift between the old and new and learn from the best practice to become champions in the new BI/DW world. Unfortunately not many people are consciously doing that when things are out of their comfort zone, and consequently their knowledge and skills become a detrimental force in the new world.

LOL, Hang, I was hoping Jeff's comment was sarcasm. I know it can be difficult to determine on these boards.

Hoops! Thanks B&L for reminding me, and sorry Jeff, for the lack of my sense of humour.

Well, it couldn't be better timing as I just told my friend that I have improved my English writing skills through the forum. Obviously it's time for me to focus on listening skills. However you see the point, switching to something new is always challenging as I originally came from non English background, but I am getting there.

So let me rephrase my previous comment as an extension to Jeff's sarcasm.

Yeah, and DBA should remain the king of DW without any skill surgery, as they have ruled the Database World for so many decades.

BTW, I have just also changed my profile so that people would not pick on my lack of humour sense.