I've got an overnight extract running through ReportCaster that I would like to cut runtime on simply by doing 5 concurrent extracts each pulling 1/5 of the records.

My output is delivered by ftp, so I don't mind transferring 5 files separately and putting everything together on the other end. Can anyone provide what they think is the simplest approach for doing this?

Instead of each of your five extracts limited to a fifth of the number of records, a slightly different approach will ensure that no duplication would occur.

Take the values of a key field in your data and divide them into 5 ranges. Then you can pass the start and end value of each range to the extract program via Caster parameters. Add WHERE statements to the extract program.

Something like this:

-DEFAULT &RANGE_FR = 0;
-DEFAULT &RANGE_TO = 10000;
-DEFAULT &RUN_NBR = 1;

-SET &DSNAME = C:\extracts\extract || &RUN_NBR || '.ftm';

FILEDEF EXTRACT DISK &DSNAME
TABLE FILE INPUT_TABLE
SUM
...
BY
...
WHERE KEY_FIELD1 FROM &RANGE_FR TO &RANGE_TO
ON TABLE HOLD AS EXTRACT FORMAT ALPHA
END
-RUN

I hope this helps.