Joe Celko's Data and Databases: Concepts in Practice
By Joe Celko
Chapter 14: Data Encoding Schemes
Chapter 14: Data Encoding Schemes
Overview
You do not put data directly into a database. You convert it into an encoding scheme first, then put the encoding into the rows of the tables. Words have to be written in an alphabet and belong to a language; measurements are expressed as numbers. We are so used to seeing words and numbers that we no longer think of them as encoding schemes. We also often fail to distinguish among the possible ways to identify (and therefore to encode) an entity or property. Do we encode the person receiving medical services or the policy that is paying for them? That might depend on whether the database is for the doctor or for the insurance company. Do we encode the first title of a song or the alternate title or both? Or should we include the music itself in a multimedia database? And should it be as an image of the sheet music or as an audio recording? Nobody teaches people how to design these encoding schemes, so they are all too often done on the fly. Where standardized encoding schemes exist, they are too often ignored in favor of some ad hoc scheme. Beginning programmers have the attitude that encoding schemes do not really matter because the computer will take care of it, so they don?t have to spend time on the design of their encoding schemes. This attitude has probably gotten worse with SQL than it was before.
Database designers think...
Copyright Morgan Kauffmann Publishers 1999 under license agreement with Books24x7