Browse Prior Art Database

A Self Correcting Segment Aging Frequency Heuristic

IP.com Disclosure Number: IPCOM000121381D
Original Publication Date: 1991-Aug-01
Included in the Prior Art Database: 2005-Apr-03
Document File: 1 page(s) / 36K

Publishing Venue

IBM

Related People

Rasmussen, EC: AUTHOR

Abstract

Described is a technique for use in a computer operating system (OS) wherein an aging heuristic is responsible for a reduction in disk access and/or CPU overhead.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 100% of the total text.

A Self Correcting Segment Aging Frequency Heuristic

      Described is a technique for use in a computer operating system
(OS) wherein an aging heuristic is responsible for a reduction in
disk access and/or CPU overhead.

      When a virtual memory OS brings an object from secondary
storage to primary storage, some other object (or objects) must leave
to make space.  Typically, the OS chooses the least recently used
(LRU) objects as victims. Many operating systems approximate object
ages by periodically enumerating the objects and marking those which
were referenced as recently used.  This is called aging.

      Unfortunately, the optimal aging frequency varies depending on
system load and application mix.  If frequency is too low, the LRU
approximation is poor.  If it is too high, CPU time is wasted.

      The technique described herein varies the aging rate to match
dynamic OS conditions.  When the system starts swapping or discarding
memory objects, the aging rate is increased.  When the swapping
stops, the aging rate is lowered again.

      This yields high-frequency aging when it is needed, and low
frequency when it is not needed.  During periods of intense swapping,
the system has the time-of-reference granularity it needs for good
LRU approximation without wasting CPU time when conditions are
stable.

      Disclosed anonymously.