Tuesday, September 23, 2008

How can BI get agile?

image We are hearing a lot about how Agile software development techniques deliver more value quicker to the end user. The market and competition is changing faster every day, software development must be able to change with it. In our world today, we can't afford to wait for results from a traditional waterfall approach, when what we deliver will not meet current requirements.

Anyone that has been exposed to agile methodologies sees why this is such a powerful approach. I have to ask myself... Why don't we see more BI projects leveraging agile? Agile has been more prevalent and quickly adopted in the SOA / App / User Interface areas than in the data side of the world.

If we think back on the history of BI projects, we are not too far from the time when we had a failure rate of 50%! Why did they fail? Long duration, high complexity, high cost, and not delivering what the business needed, when they needed it. Sounds like a recipe for Agile to me!

I challenge you to shake off the dust and find how you can deliver in new and innovative ways. The excuse that "We don't have that data in the warehouse" is simply not acceptable in today's world. We need to be able to gather requirements, profile data, data model, build ETL routines, build reports / dashboards / visualizations, and deploy to production in the space of 2 weeks. Or why not 1 week?

Here is some reading:

Agile Data - DW Best Practices

Agile Manifesto


Stephan Deblois said...

Hi Mark,

Yes, DW/BI is made for Agile. I have design BI solutions for years and turning to AgileDW was a great move. It makes so much sense and it put the power of prioritizing the product backlog into the business hands...and it gives the ETL team a frozen scope even if it is only for 4 weeks (but it makes a great difference on their productivity). You are also right about the importance of delivering much more quickly than we did in the past. Agile will help for this but most important is an architecture where everything gets automatically generated (and updated) from your DW metadata (staging tables, indexes, publishing views, fundamental star schemas, etc). Strong naming standards are also important as the foundation of a self-adapting BI environment. In other words, strong naming standards are the key to develop powerful automated DW maintenance tools.

Mark Cudmore said...

Stephan, thank you for taking the time to share. It is obvious you have had success applying agile techniques to BI. I really like your reference to "self-adapting" BI systems.


This is a personal weblog, and does not represent the thoughts, intentions, plans or strategies of my employer.