Beefing Up The Buffer Cache

Modern life demands fast computations. Even the slightest latencies can have severe consequences and cost companies a fortune in lost revenue every hour. A vital component for reducing latency in computer systems is the cache. A cache's main purpose is to store data that is frequently re-accessed, and thereby reducing the time it takes to fetch popular items, instead of constantly fetching from slower devices. In this report, we examine the page cache in the Linux kernel. We investigate how the page cache can be monitored and profiled, as well as discuss how it could be improved upon to reduce latencies in computations. The page cache is complex so we will seek to explain some of its relevant features. We introduce a method of tracing the page cache to obtain information about its current implementation, and seek to explain its behaviour. We will also feature ideas on how the page cache's replacement policy could possibly be improved.

Authors dc:creator

• Júlía Oddsdóttir 1996-
• Sandra Ósk Sigríðardóttir Bender 1989-

Contributors dc:contributor

• Háskólinn í Reykjavík

• Hugbúnaðarverkfræði
• Tölvukerfi
• Software engineering
• Computer systems

Language dc:language.iso

en

Handle dc:identifier.uri

http://hdl.handle.net/1946/31185

OAI identifier oai:identifier

oai:skemman.is:1946/31185