Imagine you have one normal-sized sheet of scratch paper to write on and a list of a million numbers. You read the list, making notes on your scratch paper as you go. Afterwards, someone asks "How many times did 1134547 appear in that list?" You look at your scratch paper and respond, "Between 15000 and 16000 times," and you are right. How can this be possible? You have kept track approximately of millions of things while only writing down thousands of digits. Programs which try to extract conclusions from huge quantities of data on the internet find themselves in scaled-up versions of this same situation. The volume of data being continually presented swamps the available memory, requiring new algorithmic ideas. We will discuss some of the problems and solutions in this field.