the task is to import invoice records from the Deutsche Telekom. The data in the source file have to be devided in several tables (such as Customer, calls data, invoice data and so on). The import takes usually 1 to 2 hours. The codeunit is just a single codeunit which opens the source file, selects a records and looks which type of record it is. Depending on the record type the data is pushed into a table. Reading a record and deviding it to the desired table is a loop, which will be finished when the footer is read. I guess it would be okay if i could make every loop “sleep” for a while, so the database can access a request from a user. It wouldn’t matter if the runtime of the codeunit would grow up to 3 or 4 hours. More important is that the users can do their work in the meantime. btw: of course we read out the source file at night (if it is available). But if we receive a file within business time it has to be imported asap and this causes the described problems.