Extended Data Lifetimes
The lifetime of data is now being extended for a number of different reasons. Users often have varied performance requirements for time-based or historical data: recent data might be accessed frequently, while older data is accessed less often. Call these conceptual access rates hot, warm, cool, and icy, respectively, ranking from most frequently accessed to least frequently accessed. Keep in mind that these data access states are largely conceptual. Cool and icy have a loose correspondence with the temperature‑based block‑level compression state of COLD and the warm and hot states loosely correspond with the identically named temperature‑based block‑level compression states.
In a warehouse with massive historical databases, the volume of data is typically inversely proportional to the data usage, as illustrated by the following graphic, where the ordinate represents the relative warmth of the data and the abscissa represents the volume of data represented by the respective measures of data warmth.
In this picture, the temperature of the data reflects its access rate. The optimal storage for hot, warm, and cool data is online disk that is directly accessible to the data warehouse.
See “CREATE TABLE” and “ALTER TABLE” in SQL Data Definition Language for information about how you can specify the way an individual table deals with block-level compression based on the temperature of the data.