I am looking to design a cube(s) for a grocery retailer. At a product level there are approximately 11million rows coming into the database per day.I have scoped out the reports that I require - At this point most things will be done weekly, by period, and quarter - all with year on year metricsMy question is how do people generally handle this much info? Do you keep 2 years of product level sales etc info aggregated to weekly level (sales, units etc) in the fact table being pulled into the cube - i.e. one massive cube!!! with all the related dimension tables OR would you do quarterly cubes and then to do year on year measures pull data into the reports from 2 cubes?I have done lots of research and can't find an answer on performance for the amount of "time" required to be stored - only around the amount of dimensions and aggregations.Thanks in advance
↧
