Relatively small, probably under 10 items. That is why I was looking for another way to do it. The hash and sorting was too much overhead for the amount of data. Although I will be processing about 100,000 files daily the data per file (epoch times) will most likely never exceed 5 or 6 items.