Data Flow->> Slow changing Dimension

Here's a brief talk of SCD.

Put two useful link addresses before you speak. The author's two papers explain what SCD is and how it is applied

Http://www.cnblogs.com/biwork/p/3363749.html

Http://www.cnblogs.com/biwork/p/3371338.html

Slow changing dimension translation comes down to the slowly changing dimension. It is applied to the loading of dimension table data in the Data Warehouse. Because data is always growing and changing, the processing scenarios for incrementally loaded data need to be processed after the first full load of data, as well as the question of whether the data needs to retain historical data. Slow changing dimension must meet these technical aspects of the problem. There is a component called slow changing dimension in SSIS, but its implementation can actually be replaced by other methods. The three common types of SCD are reference: http://www.cnblogs.com/biwork/p/3363749.html. In fact, just keep a record, keep the past history (by the Time field or the Label field to indicate a valid row), keep only the current line and the "newest old line".

To replace the slow changing dimension component:

1) Pure T-SQL Merge method to achieve

2) Use lookup in SSIS, Conditional Split, multicast and other controls to implement the SCD effect (where Lookup can also be replaced with the merge component, just to sort the data in advance)

Why replace the slow changing dimension component?

Slow changing dimension components either match the keys of the lookup table (lookup tables) or use the OLE command later to update the output lines of the change Attitude branch, insert to destination Table is also not fast load (because there is a lock conflict between insert and update, it is inserted row by line). As a result, the performance of the slow changing dimension component in the case of large data sets is certainly poor, or less than the other two methods. (from professional Microsoft SQL Server integration Services)

Data Flow->> Slow changing Dimension