DSXchange

No, not .uvconfig, dsenv.
Do the following

Well, your record size seems to be large and hence this might be the optimal performance that your might ever get. It also depends upon your network traffic.
Same goes for inserts and updates as well. What database are you accessing with the ODBC stage?

You guessed it right Ray, you are the man.

Great. I actually meant to write "job parameter", but for some reason wrote "stage variable". I edited my original post.

Transformer is the only place in a job where job parameters are recognized by thier name alone and not with hashes. Remove the hashes and the quotes.

A single stage wont be enough. But you can design a job to do this. Pass your dataset through filter stage and filter on the second column value = "ABC" . This will give you the entire record for every key that has ABC . Load into a lookup file set if data size is small else load it into a dataset (...

SO in the entire dataset, if you find ABC in col2, then change col3 value of all associated with identical first col, and the rest of the data will have zero? Did I understand you correctly?

You can stick the -warn 0 option anywhere after the -run and before the Project name. Setting warning limit for a sequence job is enough for all its subordinate jobs.

You can find different ways of checking the file size by doing a search. Keywords of search are "check for empty file". Depending upon the result, you can branch your execution.
PS: It does not have to be a script, even a BASIC routine will do the trick. Search for it.

Thanks Ray, I will give that a go. Strangely enough, our retrieval from the tables is now displaying similar levels of poor performance. Cheers, Stu My reply was to your problem stated above. If the select on a table is horribly slow, its more than likely due to back pressure cause by your target d...

Your most welcome.

It will. The dataset, when created, holds the node information as well (File Descriptor File contains this info.) Hence, you need to stick to the same configuration for creating and reading a dataset. I believe there is another post out there on the same discussion.

After making the changes, you need to stop and start the datastage engine.

You know what, dont give the second argument. By default DecimalToDecimal() takes in trunc_zero. Do the same with DecimalToDfloat(). Take out the second argument.

It does not matter what the target is. You can also stage your output to a flat file and then from there load to your database. It does not matter.