I just noticed that this question needs an answer. I found your question while just browsing around.
You say “upload SQL table to mainframe dataset.” Does “SQL table” mean SQL Server?
I don’t think you want to upload the table directly to the mainframe. The format of the table is specific to your RDBMS and I don’t think anything on your mainframe could read it. Your “SQL table” is formatted for a specific product.
First you will need to UNLOAD your table. I say UNLOAD (and not EXPORT). You will want to have your data in a flat file for FTPing to the mainframe. And the file needs to be plain data. It cannot be formatted for any specific use. That is why you cannot use an EXPORT; EXPORT processes tend to leave the data in a format usable only by the same companies IMPORT. You want to end up with records, not rows.
As an alternative, you could use a script to SELECT from the table and write to a flat file.
You will need to know the file layout of your unloaded table. If your unloaded data contains things like NULL INDICATORS, or variable length information.
Then normal FTP process to copy the file to mainframe dataset.
Then use your file layout from above to determine file layout of your new dataset, adjusting for things like variable length records.
Now you have a flat file on z/OS and you have a record layout to match.
During the FTP process, translation can be made to occur, translating from ASCII to EBCDIC.
Now, from the mainframe, you can do whatever you like.