I have a large table having millions of rows. There is one datetime column which captures the date and time a row is inserted to the table. Now, I want to copy the table's data into another server's table with same structure. Here the client wants me to migrate the data in a page by page manner (say 100 rows at a time). This table does not have any identity or unique key. I am thinking of using the row_number () function on the datetime column. But I have one doubt... say the trasaction fails somewhere in between... say at row number 115... it is written like the process will start again from 101 and not from 1. But is there any guarantee that the row_numbers will be same considering the datetime field can have duplicate values?