>>Hi everybody,
>>
>>Do you have any experience or some statistics in speed difference using WITH BUFFERING = .T. and with no buffering in Select-SQL?
>>
>>The process what normally takes ~1h now takes more than 3h and the CPU usage is at maximum level.
>>
>>Do you think we better use a different approach? The tables are table buffered and have ~300K records in each.
>
>No, I've never done a proper testing for this. I usualy use with (buffering=.t.) against my curently used and changed CAs or Buffered tables. But they are small. But 3h for 300K records? WOW!

There are lots of tables (Patients, Address, Employers, Phones, etc.) Address has 640K records. My idea was that when we receive a file from a client, we may have duplicate patients (same patient, different visits) in this file for patients, which we don't yet have in our tables. So, while we're loading records, we don't want to create the same patient (and related) information twice.

This program is written by my colleague, but we were changing it together (my idea of using WITH buffering = .T.) about two weeks ago. So far we have small files from clients, so we didn't notice a problem. Today we tried to load 15K file and then the problem showed up.

I would really like to hear some input. We're using TABLEVALIDATE = 7.

For now we switched back to the original copy of this program from VSS. I guess, we would need to change our approach and use SEEK and LOCATE instead of SELECT-SQL.