What I used it for was scheduling some batch document-publishing jobs. We have a lot of documents on our web site, and they each take a certain amount of time to publish (depending on what the templates do). We did a republish of all documents, but there are too many to do over one night, so we had to split the republishes up. We'd estimated how many documents we could publish per night. Our site has sub-sites corresponding to departments within the organization, and we needed to republish all of a subsite the same night. Each subsite has a certain number of documents (some have 100 docs, others 500, etc.), which I output using an SQL query. To optimize the schedule, I used Algorithm::BinPack to "pack" the sites into bins of, say, 3000 documents per night. It worked very well.

(Of course, it didn't go perfectly. We'd estimated 3000 per night, but after a few nights we adjusted that number; that was easy to do, just change the bin size in the script. Also it turned out that certain, more "special", subsites needed to be published on certain days, so we had to shuffle them around a bit.)