Use a GUID for the failure_id field, a GUID is, by definition, always unique. I also suggest you check message#1046937.

>I have a table equipment_failures_and_solutions.dbf
>
>A will distribuite a free software over the internet
>so many technicians can download it and add their equipment
>and failures to the table
>
>Every tech will send me his table weekly
>
>Then, I will append the tables and send back to them
>in a only large table
>
>Example of one record
>
>USER = TOM
>device = TV
>Model = 2007X
>Failure = NO PICTURE
>FAILURE_ID = TAL29X788DKK
>
>The failure_ID would prevent the same record from being appended again when I sent the large table back to them
>
>1- How to create a random code to the failure_id above ?
>
>2-How wide should be the code to prevent the creation
>of a repeated code in a table with about 10 milion records or more ?
>
>3- Is a random ID field a good idea ?
>
>Thanks
>
>Moises