An architecture choice to build reports on top of Django

[nilamo]

There's free demos available so you can try it out. I've never used a cube in a real production environment, but I've created a few and messed around with them enough to be aware of the idea of them. The thing with a cube, is it's not possible to get information about individual things. EVERYTHING is rolled up in various ways, depending on how you set it up (grouped by time period, product type, sales person, that sort of thing). "creating" a cube is really just running a wizard that runs a bunch of sum()/avg()/group by queries and storing the results so other queries are very fast.

...they're mainly used to hand to end users (they access the data via something like Excel), so they can build their own reports. Sort of if they have a different idea for how they want to look at data each day, instead of having a set report that they always go to.

It's neat, but with a little work you can display any information in a multidimensional array like a cube does. The one I played with (microsoft's version) was basically everything rolled up into one number, and then you add columns and rows to "drill down" into that number, to break it apart by month/week/day of the week/hour, and then again by department/salesperson/product type/product.

I was unimpressed, but that was probably because when someone says "OLAP cube" I was expecting it to do a whole lot more than just pre-calculate summarized data for me, and make that available to excel lol