This counter counts pages actually pulled in from disk to satisfy page faults, a subset of all faults that could not instead be resolved from memory lists already held in RAM. Administrators use it alongside the overall page-fault counter to gauge how many faults are costly hard faults requiring disk I/O rather than cheap soft faults resolved from memory. For example, if only a small fraction of a period's total page faults actually required pulling data back from disk, that indicates most of the memory pressure was absorbed cheaply rather than through costly disk access.