Forum Discussion
Microsoft EmployeeFat and Slow Golden Dataset Situation How Could Dataflows and Datamarts Help?
For sure! Another client is doing intense chaining with mixed storage model, i.e direct query an existing dataset (golden dataset) and add additional table using other data sources. It is more complex as they propragate.
Super UserHi cerebraino
A golden dataset is a good approach. But in my experience a dataset should only relate to a single question that is trying to be answered.
For example HR data should be a single dataset. Finance should be a single dataset, Sales should yet be a single dataset. This allows for it to be broken into specific business units with related tables.
With regards to datamarts right now it is still very new and I would not recommend using it for production loads. It will get better and hopefully have more options as time goes on.
Also make sure that each dataset is designed using a star schema with 1 fact table per business area and directly related dimension tables
- cerebraino
Thanks for the suggestion about having different datasets for each "business area", instead of one , however my boss insists in just having one for all possible scenarios and questions, I will try to convince him to "split the sacred cow", thanks again! Juan
- GilbertQ
Yeah I know it might be diffucult now, but it will save hours of time, effort and slow reports.
It can always be done, it is often just getting the time to get it done!
