- Disc is very much slower than CPU
- Seek time is much larger than read time
- An inefficient job that ‘wastes’ CPU may give better overall response time than an efficient job that leaves the CPU idle most of the time.
- the table is split over many discs
- different discs can be accessed in parallel with minimal contention
- there are multiple CPUs in use so that there is no contention for CPU (although PQO can help even on single CPU systems).
For example :