Cool and Icy Data
Typically, there is a vast quantity of cool historical data in the data warehouse that is accessed lightly and a lesser volume of hot and warm data that is accessed frequently.
Note: The distinction between cool and icy data is conceptual and has no direct parallel to temperature‑based block‑level compression.
While cool data might be accessed lightly on average, it still has temporal hot spots, such as those that occur when performing comparative analyses of sales data between different time periods. Similarly, if the relational schema is modified by adding new columns or indexes, or if column data types are changed, then the affected data cannot be considered to be dormant. Finally, if data is periodically accessed and somehow recast to make historical data relevant within the current business context, then it is not dormant.
Truly dormant data is rarely, if ever, accessed, and is typically retained for various national and international regulatory reasons such as the Sarbanes-Oxley Act in the United States, Bill 198 in Canada, or the Eighth Company Law Directive 1984/253/EEC in the European Union rather than for operational reasons. As a result, the data stored in the warehouse is frequently a mix of both important and unimportant data (Unimportant from the perspective of the day‑to‑day operation of running the enterprise, not from a legal perspective), and a flexible management system is required to allocate the appropriate availability, reliability, and privacy levels for the data as its usage changes across time.
Icy, or truly dormant, data is a good candidate for alternative storage such as tape or optical disk. The treatment of icy data is not the subject of this chapter.