This read-only integer reports the number of major page faults per second system-wide, faults that require transferring a page in from a hard disk rather than from memory. An admin treats a rising major-fault rate as a strong indicator of memory pressure actually causing disk I/O, far more costly to performance than minor faults. For example, a spike in major faults coinciding with sluggish application response confirms to an admin that the system is disk-swapping data that should have stayed in RAM.