Actuaries base most of their work on historical or projected data, stored in flat files or Microsoft Excel spreadsheets. While these files generally are of relatively small size, when it comes to individual GLM pricing they can be quite huge as they contain the entire insurers' accounting and contractual individual historical records. This is more and more the case in our "big data" new world.
Under these conditions, getting the best performances out of hard drives, both when reading and writing such files, is crucial so that actuaries get their results within an acceptable timeframe. This paper points out some ideas in order to get the most out of our "hard drives".