One approach is a "bulldozer" approach - take as many data as possible, at as many locations as possible and as frequent as possible. From the first glance this seems to be a bullet proof method - you will never miss anything. In fact, the opposite is true. The amount of data collected quickly becomes overwhelming and unmanageable, and apart from the difficulty of retrieving necessary piece of information, it presents another unexpected challenge. The value of storing the data is in ability to keep track of historical records, because only a relatively long period of time can be representative for the actual performance of any complex system, performance building included. Now, an attempt to estimate what would it take to create a data storage for one such building, utilizing the "bulldozer" approach described above, hits an obstacle - the data warehouse for keeping track of ALL data will cost over $400,000 !! No wonder, it is decided, or rather occurs automatically, that the wast amount of accumulated data is discarded after a relatively short period of time to let the room for the new batch of data. But what about the analysis?
What if we need the data for more than one month, and typically we want to monitor performance for at least a year? How can we even be assured that what we need is there, when what is collected spills over?
Following up on and consistent with what I have discussed previously (see e.g. Lessons Learned - Part 2) there is a need in the agreed upon hierarchy of the data sets, common format of data being collected, stored and retrieved. With the time it may and probably will evolve into the industry standard. But the work needs to be started, or we are going to face the hurdle not unlike or even worse than the Tower of Babylon - not only being unable to speak one language, but not even understand ourselves...
In order to be able to communicate we need to speak one language.