Log-structured File Systems In the early 90’s, a group at Berkeley led by Professor John Ousterhout

known as the log-structured file system [RO91]. Their motivation to do so was based on the following observations: • Memory sizes were growing: As memory got bigger, more data could be cached in memory. As more data is cached, disk traffic would increasingly consist of writes, as reads would be serviced in the cache. Thus, file system performance would largely be deter-mined by its performance for writes. • There was a large and growing gap between random I/O perfor-mance and sequential I/O performance: Transfer bandwidth in-creases roughly 50%-100 % every year; seek and rotational delay costs decrease much more slowly, maybe at 5%-10 % per year [P98]. Thus, if one is able to use disks in a sequential manner, one gets a huge performance advantage, which grows over time. • Existing file systems performpoorly onmany commonworkloads: For example, FFS [MJLF84] would perform a large number of write