<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Robert Arkiletian wrote:
<blockquote
cite="mida18d8e260710050905t71b5338fo864c71c582897eeb@mail.gmail.com"
type="cite">
<pre wrap="">On 10/5/07, "Terrell Prudé Jr." <a class="moz-txt-link-rfc2396E" href="mailto:microman@cmosnetworks.com"><microman@cmosnetworks.com></a> wrote:
</pre>
<blockquote type="cite">
<pre wrap=""> I just ran across this today over at <a class="moz-txt-link-abbreviated" href="http://www.linuxtoday.com">www.linuxtoday.com</a>. I tried out the
following two settings on my K12LTSP 4.2EL box to see if they would indeed
speed up response time for desktop apps.
/sbin/sysctl -w vm.swappiness=1 (default is 60)
/sbin/sysctl -w vm.vfs_cache_pressure=50 (default is 100)
These are, as you might've guessed from the "vm" in the name, virtual
memory settings for Linux (the kernel).
</pre>
</blockquote>
<pre wrap=""><!---->
Will both these changes only make a difference if your server is
actively using swap? IOW if you never hit swap, do they matter?
</pre>
</blockquote>
<br>
My server at home doesn't swap. However, I still notice a remarkable
increase in app response time. Now, on a busy LTSP server with, say,
40 users, it'd probably benefit even more.<br>
<br>
--TP
</body>
</html>