It's a real SQL Server database, but the part that will be intensive will be taking thousands of rows for each of over 4,000 objects, making mathematically different copies that get stored back to the database (same table, different "type"), then doing that once more before comparing each of these final objects to each other. That's part one. The next part will first create a similar amount of data (but for about a dozen different types of objects). These will be combined in a few different ways (some are still TBD), then each of these will also get compared to each of the original 4,000 objects! If this phase produces results, the next one will have millions of rows for many of those objects.
Funny you should mention about spreadsheets and flat files not being databases. I once had a manager who "thought" he knew what he was talking about (LMAO!) and he would definitely call any collection of data a database!