A number of parallels can be drawn between putting together an IT solution and a real-life building construction project. In both cases "things" are architected, designed, constructed, and invariably, the clients are neither happy with the result, nor with the "builders" nor with the process.
If we can take the analogy one step further - there is a reality show called `The Block` where people move into a dilapidated half-completed building with the task of completing the building and decorating it with the aim of selling it.
Working with the "givens"
As business intelligence consultants, our heads are often put on "the block". We are sometimes contracted to make a business solution work successfully with a lot of "givens". We are dropped into the deep end to create a business intelligence solution from an empty shell which is still far from an operation data warehouse. In these environments you often have to be content with second best and have to deal with very frustrating limitations as you have to run around unnecessary restrictions.
We recently successfully completed exactly such a project for a cellular telephone operator in central Africa. In this case, the foundations were already laid - the SAN was already in place, set up and configured. We weren`t allowed to touch it (in actual fact, we had difficulty inspecting its configuration). The walls were already built - that is, the DBMS was already in place and configured. Again, we had no say about it, but we had to fit the contents of the proposed data warehouse into that framework. Imagine a building that was built as a bachelor`s pad, and you have to fit a four-member family`s furniture and household items in there. We could do minor cosmetic changes (ie performance tuning from a data modelling perspective), but on the serious configuration issues we had to settle for a compromise. This is extremely difficult, especially if you know from experience that there is a better way.
What to do?
In this particular case, it was a major landmark project. How does a builder walk away from completing a prestigious landmark building?
In such a situation, your effort estimations and project management have to be spot-on. It is difficult enough to find project managers that understand and can manage the intricacies of frequently changing iterative short-cycle business intelligence projects, never mind handle unknowns that are given. This is a crucial point - the "givens" are often "unknowns". For example, a pre-set disk configuration may work well for small data volumes during development, but during stress testing the performance drops below the acceptable, and you cannot tune it any better with the limited facilities at your disposal.
To relate to the analogy - you cannot really adjust the walls to cater for more furniture, if you`re not allowed to touch the foundations.
Your project manager has to have hair on his teeth. Problem handling becomes the activity of the day - often at crucial stages of the project, such as during stress testing a few days before the already-scheduled user acceptance testing is about to start. SLAs with the other parties - such as the system managers - have to be in place, and may have to be pushed to the boundaries. There is no way in such an environment that you can let an issue ride, and hope to correct it in a subsequent phase. The subsequent phase may jump its own surprises on you!
All is well that ends well
Ideally, the business requirements should first be determined, and then the solution be architected to address the requirements. Technology should not drive or dictate the solution. This results in an on-going battle to successfully complete the business intelligence solution in an often way less than optimal situation - especially where technological components are pre-specified, or even worse, already in place.
Reverting back to our analogy, although they curse the builders all the way, eventually the people move into the completed house. In our case, by using on-the-button project management with extensive problem handling - which at times irritated the sponsors to no end - we managed to complete the solution and walk away with a signed sign-off certificate, on time and on budget.
Share