Hey guys,

Firstly this is a copy and paste from gpforums so sorry anyone who checks both.

I'm just trying to decide on the best way to support a "database within database" concept. Basically the situation is we have a web application where users can upload big as spreadsheets. They can use a form to enter some data and an AJAX runs using one of the columns from their spreadsheet to match and if it matches, populate some other fields with the associated row data.

In terms of numbers we are expecting spreadsheets of up to 20 columns and maybe 30,000 rows. We could have a few hundred of them.

Our (ex) developer performed some "tests" and decided EAV would be best. Well his test must have been 3 columns and 10 rows as EAV seems to such with these volumes no matter what we do with indexes. Like almost 5 minutes to load a spreadsheet some times which obviusly won't work for AJAX. The ideas we have come up with:

- Find a magical solution to make EAV work
- Have one table per spreadsheet (leaning towards this atm)
- Look at no-sql DB (scary)
- Maybe store the databases as files on the filesystem and use PHP to parse and "query" them (probably bad idea)

What are people's thoughts on this? We are using MySQL 5 for the DB server but could implement something else.