Monday, July 26, 2010

System Center Operations Manager and BI

Last week as part of a presentation to the CIO summit in Auckland I had a bit of fun criticising conventional corporate BI efforts as being too focussed on data management. Basically, the argument went that companies spend too much of the budget worrying about getting the data right and not about enabling people to use it – and that ‘easy to use’ BI front ends are a pointless exercise. It’s not a hard argument to make – you can expend a great deal of time and effort on staging, transforming, cubing, KPI development, and reconciliation quality checks – usually driven by the fact that corporate BI is focussed on financial measures which drives people to want the data correct to the cent.

The natural question that comes back in response is what should be done instead?

Operations management using a model driven tool like System Center Operations Manager provides some useful pointers.

Think about it – what’s operations management?

= Data + Model + Health logic + Alerting for response

What’s BI?

= Data + Model + Business Health Logic + “Strategy execution” for response

What’s smart about model driven operations management?

The tools (like SCOM) monitor systems that can produce huge volumes of data; they give that data context with hierarchical models (à la cubes) that can even be graphical (LiveMaps/Visio 2010 diagrams); they provide health indicators (cf KPIs) and they empower end users by making the information available for more detailed analysis at the desktop.

So how would you put this kind of model into action for BI?

How about mirrored true to source data sources using cheap databases (maybe SQL Express/MySQL). You’d need a grooming schedule to stay on top of the data volumes – perhaps and agent model just like the operations management tools have.

To collect the data from multiple sources, and give some context a cube model is attractive, but what about implementing with PowerPivot. It puts the end user in charge. They can create health indications in an Excel like environment; and they can submit back into a shared (Sharepoint) repository.

I’ve got to present at the longwinded FST 3rd Annual Technology & Innovation – the Future of Banking & Financial Services New Zealand conference in August. I’ll be exploring this in more detail and hopefully have a few examples to test out.

No comments: