My first Publication Agile-Data-Warehouse-Design-eBook

242 Chapter 8 How Many Figure 8-9 ORDER FACT Accumulating Snapshot Staged accumulating snapshot ETL processing will need to be streamlined if real-time DW/BI requirements exist For real-time DW/BI, the latency introduced by staging each milestone in its own fact table first may prevent an accumulating snapshot being updated urgently enough for current day reporting requirements. If streamlining the ETL process becomes paramount and the milestone fact tables are not needed for queries, they can become un-indexed staging tables that are truncated at the end of every load cycle or be replaced by ETL processes that act as virtual tables, piping their inserts or updates directly to the inputs of the accumulating snapshot process. If a real- time snapshot and queryable detailed fact tables are required, the staging tables can be implemented as un-indexed real-time partitions (covered shortly) that are fully indexed and merged with their fact tables by conventional overnight ETL. Fact Types Additivity describes how easy or difficult it is to sum up a fact and get meaningful results If the most important property of a fact table is its granularity, the most important property of a fact is its additivity—which tells you whether or not its values can be summed to produce meaningful answers. This is important because stakeholders almost never want to see individual fact values. Instead they want to summarize them, and the easiest way to do that is to sum them. Facts are divided in three types based on their additivity: Fully additive, non-additive and semi-additive.

My first Publication Agile-Data-Warehouse-Design-eBook | Page 262