Originally Posted by jiml8

T
Presently, I am working on a contract where I am refactoring a spreadsheet (a really HUGE Excel spreadsheet) into a database and a compiled program. The spreadsheet is the largest and most complicated one I have ever seen, and I am seriously impressed that the client has made it work. However, they have gone much further in this direction than they should have; recalculations are taking over 12 hours and they are completely filling a 4 Gig Windows machine with the spreadsheet - to the point where a number of their cases have to be broken up and run separately, then recombined at the end.

So I am reorganizing it into what it should be. After reviewing their current capability, their spreadsheet, and all their requirements, I proposed an approach that would have me designing a database and a computation module that would replace all of the data storage and computation work that was done in the spreadsheet, then outputting all results TO the spreadsheet for post-processing and final reporting.

The reasons I chose to use the spreadsheet for the backend were several: the users were already familiar with it, it provides them a great deal of flexibility to post-process their data as they change their model, and the reporting capability of Excel is extensive and they could continue to use their extensive set of pivot tables to focus on parts of the data that was calculated, and they could easily and conveniently add to, modify, or replace portions of their output reporting as they saw fit without involving their IT department to make the changes.

Originally Posted by jiml8

The essential point, though, is that they were able to get this far in Excel. They couldn't have done it with any other spreadsheet. That they went TOO far is indisputable, but these things happen and they are working to correct it now. Without Excel, they wouldn't have made it.