- za že The world of database design will be explored, step by step, in this 14-part series, Process-Driven Data Design. by Frank Sweet Have you ever noticed that there are just two kinds of data entities in the real world? Imagine, say, a 10,000-record vendor file where, on average, 100 new vendors are added each month and 100 inactive ones purged. Its monthly volatility (turnover di- vided by population) is 1%. Now consider a file of 1,000 purchase orders where 1,000 orders are added each month and the same number closed. Its volatility would be 100%. Compute the volatility of the per- ;nanent files in your dp shop and you'll un- cover something strange; there are two populations out there—two classes of rec- | ords. A histogram shows two peaks: one i around 1% like our vendor example, the i other near 100% like the purchase orders. ' Volatility is binary, it seems, and almost ev- » ery record falls into one group or the other. | Why is this? Does it apply to every data- 1 base? Does it reveal an important underly- ing truth about data? The answer to the last two guestions is yes. But to find the why, we will journey into the world of database design. We will , see that objectively good designs exist, and | we'il learn to recognize their traits. We'll review eguations in data flow dynamics like the above, and mnemonics like K