Hi Hosen, I suspect that the second solution will be significantly faster because your overall process (write once, read once each record) only marginally builds on the advantages of a data base, while appending data to 100 files is very fast. But I'll be very interested to read your update on this.