Since we introduced Aggregate Tables in mondrian-2.1, I have been hearing two things about them from everyone: they are really good at speeding up your Mondrian-based data warehouse, and they are really difficult to design.
So, I was delighted to come across a book all about aggregate tables. (More so, because I came across it the traditional way, by browsing the computer section of my local bookstore. I've posted the Amazon link below for your convenience, but if you cherish your local bookstore, you know what you have to do...)
The book is "Mastering Data Warehouse Aggregates: Solutions for Star Schema Performance", by Christopher Adamson, described as "the first book to provide in-depth coverage of star schema aggregates used in dimensional modeling-from selection and design, to loading and usage, to specific tasks and deliverables for implementation projects".