[66425] in North American Network Operators' Group
RE: GSR, 7600, Juniper M?, oh my!
daemon@ATHENA.MIT.EDU (David Schwartz)
Tue Jan 13 06:01:34 2004
From: "David Schwartz" <davids@webmaster.com>
To: "Richard A Steenbergen" <ras@e-gerbil.net>,
"Jesper Skriver" <jesper@skriver.dk>,
"Tarko Tikan" <tarko@lanparty.ee>, <nanog@merit.edu>
Date: Tue, 13 Jan 2004 03:00:17 -0800
In-Reply-To: <20040113031334.GH97351@overlord.e-gerbil.net>
Errors-To: owner-nanog-outgoing@merit.edu
> On Tue, Jan 13, 2004 at 02:01:48AM +0100, Jesper Skriver wrote:
> That would depend what is causing the CPU usage. If it is software based
> IP header lookups, you're not going to get any more peformance out of it
> by trying to do more lookups than your CPU can handle.
Surprisingly, that's usually not true. The cost of the IP header lookup is
generally much less than the cost of a task switch. So when the system is at
100%, it's probably doing this:
IP header lookup, wait for work / task switch, IP header lookup, wait for
work / task switch, repeat
As the load increases, it will start doing two IP header lookups before
each task switch or yield. Then three. Then four.
DS
