data warehouse - Most common granularities in datawarehouse designs -
i have been looking answer of question while:
when asking granularity, immediate examples given are: transaction, day, week, month etc. couldn't find other type of example. instance, consider 'city', 'state' etc. granularity? when, example, consider sales nationwide company? in other words, granularity of type of time?
no, granularity not related time. lowest granularity kind of transaction. 1 of examples kimball uses retail setting: lowest granularity relating product sales might item being scanned @ check-out. 2 such transactions happen @ same moment, not time-based granularity.
just could granularity of table, kimball advises working lowest granularity far more flexible - can slice , dice data in more ways. might choose have aggregated tables sum data week level, or state level, or pretty else (possibly performance reasons, or make easier users) - these unlikely lowest granularity.
using state example - presumably have lower level information within same hierarchy analyse sales data by, county, city, zip code. may have data on individual customer, specific order reference, shop or sales office involved, employees involved in processing order, etc. odd choose use state granularity of fact table, unless had specific reason aggregate transaction fact table based on order item.
where see date or time fields granularity of table in periodic snapshot facts, again these aggregated other, lower-granularity data sources.
Comments
Post a Comment