Code/Database Design Recomendations

[Tracer]

I have a large enterprise database which i must use ODBC to connect to. I need to:

1) query the enterprise and select 10,000+ complicated records from multiple tables with multiple sub-selects - This process takes anywhere from 15-30 minutes.

2) query a MySQL database for selection criteria for the enterprise records (about 1000 10 digit numbers)

3) Select from the ODBC query, the matching MySQL entries and tag/alter them based upon content

4) Group & Sort the results, print a report, dumb everything except the totals and add the totals to a MySQL DB.

Anyone have ideas on the best way to perform these functions? Should I use mySQL TEMPORARY Tables? a permenanet MySQl table which changes on each report run (erase all the content on completion)? FLAT Files? Just parse it all through GIANT php Arrays?

[AstroTeg]

Temp tables would be fast. The trick is how much data are you're working with and how much ram the machine has available. If you have too much data, then probably just a regular table will do the job (maybe use a temp name for the table though).

[Tracer]

It's a LOT of Data, each "enterprise" record will have about 40-50 fields. From Text to Int to VarChar.

THen I will have to sort/group and add /update fields using multiple selects from a MySQl database of about 1000 records