Albert,

I think the problem you're running into is "memo bloat." Because it's difficult to efficiently store variable length data (in this case the memo file), VFP doesn't reuse the space in the memo field, it just tacks it on to the end. So if myfile.exe is a 1 meg file, this will create a 100 meg+ memo file:

APPEND BLANK
FOR i=1 TO 100
  APPEND MEMO MyMemo FROM myfile.exe
NEXT

Notice that this happens even though only one record is added to the file. You may want to look for another way to store the data (files named with the table's key, perhaps) or run PACK or PACK MEMO periodically. Of course, you need exclusive use to do this so it may not be practical for your needs.

>I have a VFP table that is accessed through the VFP ODBC driver.
>
>The table consists of an Integer Primary Key and a Binary Memo data block. This table is updated with sixty transactions per minute, 24 hours a day. Records are inserted until the the record count excedes a specified capacity limit. After the capacity limit is reached, the records with the lowest Keys are updated with the new data.
>
>The problem is that update always allocates new space in the memo file. This causes the file to grow at a rate of 30k per minute.
>
>I would like to create an insert trigger that uses Seek and Replace instead of Update. The problem is that the memo file would still grow if there is no buffering enabled.
>
>What is the buffering mode durring an ODBC connection? Does any one have experience with these issues?