Control of the memory hierarchy of a system as a whole, or control of allocation at a fixed level within the memory hierarchy. In the former case information stored within the system is shuttled between one realization of memory and another, the objective being to maintain maximum hit rate in each form of memory. This movement may be controlled by

1. (a) voluntary user action, e.g. copying a file from disk to memory in order to edit it;

2. (b) system software, e.g. transfer of a page between swapping device and memory when a page fault occurs;

3. (c) system hardware, e.g. movement of a set of words from memory to cache when a word within the set is accessed.

At a given level of the hierarchy the operating system will control what fraction of that level is to be allocated to each process. This can clearly only occur where control is by system software, and refers most particularly to the allocation of memory to a process, or to the allocation of space on the swapping device. Movement between disk and magnetic tape is often separately treated as archiving. See also storage protection.