Stuff it
Recently, in Ralph Kimball's column in Intelligent Enterprise, he wrote about the potential of RFID and smart dust (computers and sensors in very, very small packages) to significantly alter the face of data warehousing because of the sheer quantity of information collected and the introduction of many new dimensions for analysis. He believes that we will be able collect 10 to 100 times more data than is the current practice and want to analyze this in a myriad of new ways.
I'm wondering if this explosion of information being collected in the retail and consumer goods sector will inspire some of the practices from manufacturing to become more widespread in data warehousing? Large industrial plants already have tens of thousands of sensors collecting data from the processes they are monitoring. This data is often recorded in a process historian (like the one provided by OSIsoft) and is compressed in a way that eliminates the many redundant readings that are provided by a process that is stable. These systems allow the collection of tremendous amounts of data, without having to store all of the non-important values.
Maybe the individual readings from RFID tags will provide more entropy and be less subject to compression? I'm not sure but it stands to reason that a pallet of razor blades is going to all go through the same state changes at the same time in a predictable fashion in most cases. This predictability of the process is a reduction in the information conveyed by each individual reading and might be subject to compression. Anyway, it seems like an interesting avenue of pursuit for data warehousing vendors.