I wish more people actually took steps to optimize disk io though; my current source tree may be in cache, but my logs certainly aren't. Nor are my /usr/share/

I wish more people actually took steps to optimize disk io though; my current source tree may be in cache, but my logs certainly aren't. Nor are my /usr/share/docs/, /usr/includes/, or my old projects. Chris Mason of btrfs fame did some proof of concept work for walking and reading trees in on-disk order, showing some pretty spectacular potential gains: https://oss.oracle.com/~mason/acp/ https://oss.oracle.com/~mason/acp/ Tooling to do your own testing: https://oss.oracle.com/~mason/seekwatcher/ https://oss.oracle.com/~mason/seekwatcher/