Data Flow->> Merge

The merge component has a similar effect to union all, which is to combine the result set of two input sources into one. But the difference is:

1) The result set of the merge input needs to be sorted first (this is doubtful)

2) Merge is explicitly required for the data type of the two result sets entered, and must be the same, for example a field of a result set is int, and the corresponding field of the B result set must also be int. However, for data lengths, such as varchar, there is no requirement that the same length be required. But just can't one is varchar, one is nvarchar.

3) Union all supports more than two input result sets and the merge supports only two.

If you look at it this way, it seems to be all done with union.

The above mentioned 1th said that the data set must first be ranked, after the experiment found that it is not. Two tables as input, although the merge requires you to change the issorted to True, then set the SortKeyPosition value of one or some of the fields in the output columns, But it does not mean that setting these two values is equal to helping me to sort the data sets when run-time, because I have observed the order of data insertions in the input table, or the Order of table +b tables, and my sortkeyposition value is set to sort the other fields. If it really helped me out of order, it wouldn't have happened. The +b Table of table A is inserted in this order. And the result is that the package ran successfully. Therefore, the result set of the merge input needs to be sorted before this is not the premise.

Data Flow->> Merge