DSXchange

Re-import the table definition for that particular table. This happened to me a couple of times as my dba changed his db script for the x.next version of the database on friday and sent out the mail on monday.

Darn, i thought id sly by without you noticing it. But you got me
The OP should really come back and let us know if it worked or not. I dont have access to DS at the moment and thus cannot test out my routine

Are you calling it by hashed file name or the VOC entry name. You need to put in the VOC entry name and not the hashed file name.

Looking at the two datasets i presume that CC_ID is the key. Then your hashed file wont look like how you showed. Hashed file removes duplicates, retaining the last identical key. If you want the opposite, then make the contents of your hashed file as source, the other file as lookup. This way you w...

Mine is 4 characters at a time

If you want to do it from designer, you can set it project wide by logging into the director, go to tools, options and set the warning limit to 1 over there.
For command line, pass warn 1 along with the dsjob -run command.

Actually if you look at my routine, its fairly simple. Its not working because its not complete. You need an 'Ans' line. I left that upto you so that you can decide what delimiter should come between 'Part' and 'RestOfData'. Desc=TRIM(Arg1) RestOfData='' Part='' EndForLoop=LEN(Desc) ...

I revisited my routine above and found some minor logical errors in it. I fixed it. You can use it now. Try to test it for different scenarios. I, currently do not have access to Datastage and hence cannot test it myself.

'Performance' is a relative word. Relative to your environment, record size, depth of transformations, network traffic, database utilization. For millions of records, a logged activity is not advisable. Use the load method. Its a good thing that your data is in a flat file. Invoke the Ora bulk loade...

If you just want to collect both the files then why not use a link collector. Merge stage is slower than the link collector. I usually tend to stay away from the merge stage.

Even i am interested to know why exactly you want to do that via routine or unix?
Is this really your requirement or just for knowledge sake?

Union All is the best way as advised by others. Its a simple sql that you need to write and get it done. Are the two tables in different databases? If yes then use the funnel stage. You can also create two files and combine them at the OS level.

Why dont you build a lookup set from the 50 records. Have the DB2 table as your input and do a lookup. This way you will get all the records from the table that have a hit on the dataset. I am sure you must have thought of that but whats stopping you from using this method. I am not very familiar wi...

Try to google the keyword ORA-12154. Try to lookinto it. Try this link for a start
http://www.is.mcgill.ca/minerva/Help/ora12154.htm

Load the 60 records into a temp table and utilize the power of UDB by doing a join at the database level. That would be faster anyday, keeping in mind the network traffic that will be induced if 30M records are to be loaded to a file and then doing a lookup.