Howdy All!
I have an ugly flat-file import that I'm working through -- EBCDIC flat file, coming straight as a mainframe tape record.
I've got the control flow working properly up to the Data Flow... I can go, download and unzip the file all in the SSIS Package (good!)
This means that the file is hierarchical -- there are ten different row types, and it's got packed decimals scattered all throughout the file (where it starts getting ugly).
Fortunately, each row always starts with an indicator to let you know what type of record it is.
So Step 1 - Flat File Source with two columns -- column 1 is the first two bytes, which is the record type; I'm saving column 2 as a dt_byte array.
Step 2 - Script component with two inputs, and ten output sets
Step 3 tier -- output destinations from script component.
When I did my initial testing with flat file destinations all seemed to work well, however as I changed over to OLEDB Destinations, I noted that I was writing out dependent tables, before writing out the root record.
I know one alternative is to disable FK constraints while writing the outputs, but it would seem that I'd have an even better solution if I could say "wait for this output to complete before writing the other rows"
Does anyone have any suggestions on the best way to control the order the outputs flow?