How many records are/expected to be in the disp_data table?

>Hey Folks,
>
>I have a situation where I need to update a table in a Database with info from another file (SDF text file, can convert to DBF or not as needed). The table to be updated (disp_data) is structured:
>
>phone, C, 10
>disp_code, C, 1
>disp_date, C, 10
>
>The file it comes from (dialer) will have this structure:
>
>phone, C, 10
>disp_code, C, 1
>disp_date, C, 6
>
>... as well as other info not pertinent to the issue at hand.
>
>The difficult part arises from the update criteria. I need to ...
>
>1) Update recs in disp_data that have matches in dialer, meaning that the recs in disp_data need to have the code and date overwritten,
>
>2) Add recs to disp_data that are in dialer but not in disp_data already, and
>
>3) Ignore recs in disp_data that are in it but not in dialer, meaning do not overwrite the code and date.
>
>Normally, we're looking at 100,000 to 1,000,000 recs in the dialer file to be processed per update session (usually once per week).
>
>I could use a SCAN loop, checking for each phone # and either adding or updating as needed, but that takes way too long, many hours for that quantity of recs.
>
>I could use three SQL SELECTs, UNION the results, then overwrite and reindex the original disp_data table when I'm done, but I could see that taking forever as well as the disp_data table grows in size.
>
>The best idea I've come up with so far (at least I think it is, for the moment) is to SELECT the matching recs, then REPLACE or UPDATE in disp_data with the result set, and SELECT the new recs, then APPEND them. However, this seems kind of kludgy to me, and I was hoping someone else had some better ideas?
>
>Any takers? The provider of the best suggestion will be welcome to a beer or two anytime you come to Phoenix ...
>
>TIA.
>
>--
>L8R
>
>Dave