5

As you can see in the image below, I have a table in SQL Server that I am filling via a flat file source. There are two columns in the destination table that I want to update based on the logic listed below:

  • SessionID - all rows from the first CSV import will have a value of 1; the second import will have a value of 2, and so on.
  • TimeCreated - datetime value of when the CSV imports happened.

I don't need help with how to write the TSQL code to get this done. Instead, I would like someone to suggest a method to implement this as a Data Flow task within SSIS.

enter image description here

Thank you in advance for your thoughts.

Edit 11/29/2012

Since all answers so far suggested taking care of this on the SQL Server side, I wanted to show you what I had initially tried doing (see image below), but it did not work. The trigger did not fire in SQL Server after SSIS inserted the data into the destination table.

If any of you can explain why the trigger did not fire, that would be great.

enter image description here

sion_corn
  • 3,043
  • 8
  • 39
  • 65

3 Answers3

6

If you are able to modify the destination table, you could make the default values for SessionID and TimeCreated do all the work for you. SessionID would be an auto-incremental integer while the default value for TimeCreated would be getdate() or gettime() depending on the data type.

Now, if you truly need it the values to be created as part of your workflow, you can use variables for each.

SessionID would be a package variable which is set by an Execute SQL Task. Just reference the variable in your result set and have your SQL determine the next number to use. There are potential concurrency issues with this, though.

TimeCreated is easily done by creating a Derived Column in your data flow based on the system variable StartTime.

Ken Rachynski
  • 320
  • 1
  • 9
  • I initially tried doing this through SQL Server. I made an edit to the initial post explaining my process. I will try implementing your SSIS solution - I will let you know how this goes. – sion_corn Nov 29 '12 at 16:39
3

You can use a Derived Column to fill the TimeCreated column, if you want the time of the data flow to happen, you just use the date and time function to get the current datetime. If you want a common timestamp for the whole package (all files) you can use the system variable @[System::StartTime] (or whatitwascalled).

For the CSV looping (i guess), you use a foreach loop container, and map an iterative value to a user variable that you map in the derived column for SessionID as mentioned above.

cairnz
  • 3,917
  • 1
  • 18
  • 21
-1

First, I'd better do it on SQL Server side :)
But if you don't want or cannot to do it on server side you can use this approach:

It is obvious that you need to store SessionID somewhere you can create a txt file for that or better some settings table in SQL Server or there can be other approaches.

To add columns SessionID and TimeCreated to OLE Destination you can use Derived columns

Igor Borisenko
  • 3,806
  • 3
  • 34
  • 49
  • Thanks for your input Igor. I can't have SessionID stored in a text file, because it has a conditional value. – sion_corn Nov 29 '12 at 16:44