Reducing Thrashing With Time-Outs
Original Publication Date: 1986-Feb-01
Included in the Prior Art Database: 2005-Mar-08
Thrashing is a common problem on systems with overcommitted storage. The more often virtual areas are moved in and out of storage, the worse system throughput becomes. Thrashing can be reduced by setting a timer after execution of a supervisor call (e.g., WAIT or CALL). The system storage manager is prevented from using the real storage associated with the related program until the timer expires. Statistics have shown that the longer a wait has been outstanding, the longer until the wait will be satisfied. Typically, systems have not differentiated between waits of various lengths when selecting virtual areas to remove from storage.