>I agree it is probably not worth the effort at this point - if you have that much data in a table it really should be in something else like SQL Server. I guess the advantage to it would be that it could save you a lot of work in certain situations. Moving a DBF to SQL Server is easy enough -- but depending upon how the app was written there can be a lot of code that would need to be reworked. ... and sometimes splitting a table or archiving records can be a daunting task depending upon how things are written and designed. Still impressive if he can do it though :)

Actually I've already hit the limit, but not with my tables. I wrote an email table, which stores only some things from the header (sender, recipient, datetime, title, folder) and then also the offset and length, so when I need to use that email, it's a fairly simple combination of fopen(), fseek(), fread(). But I've hit the 2g limit with fseek() - some of email folders contain files with over 30000 emails, many of which contain attachments (stored in base64, which is 4/3 of the original length) and they are over 2G. Luckily, I can manage by not being interested in those files :), but just wanted to report here what happened.