How does one reflect the historical changes of the primary data? [From wkingchen's blog]

Reason for forwarding: The first time I came across this idea, it was also a practical application of time dependencies. The idea was very unique.

 

Sometimes, users make some changes to the primary data (such as customers and materials), such as modifying the region of the customer, and modifying the category of the materials. So how can we reflect the historical truth in the report for this modified primary data (slow change dimension in DW?

First, let's look at the updates to the master data under the Standard settings. Assume that 0 customer has a property of 0 region, which is time-indepent. In this case, when the user updates the region of a customer, the original region will be overwritten. Therefore, the report obtains the latest region about the customer. Drilldown by region, all historical data of this customer will be included in the latest region. In this way, it is clear that the actual situation of history cannot be reflected.

At this time, we will think of setting 0region to Time-Dependent. In this case, the region value of this customer will be bound to time. When you make a report, you can obtain the original region or a new region based on the specified key date. When drilldown by region, the customer's data may be classified into new or old region, which does not reflect the historical truth.

So what is the solution? It is easy to put region into characteristic of infocube, and read the primary data to obtain its current value. In this way, the historical data of region is saved when the number is extracted. However, this is also an issue. When you need to repeat the system, the historical data of region will be overwritten by the new data. In this case, if the time-depent setting is used, you can put region into _ characteristic and specify the key date when pumping the number, which can perfectly solve this problem.

Looking forward to an unknown journey with hope.