Hello Claudio, On Fri, 17.03.2006 at 10:28:31 +0100, Claudio Jeker <[EMAIL PROTECTED]> wrote: > Finally packet forwarding should be mostly unaffected by userland CPU > usage as it runs with much higher priority in the kernel.
I think this is actually not the case in real life. When one of my routers reloads it's session (eg. the remote end reboots), this consumes enough CPU to cause serious packet loss. Although I didn't really measure it, from the few cases I witnessed it, I think packet loss is in the order of 20-50% while the reload takes place, ie, for one or two minutes (1 full table plus a few other routes). Best, --Toni++
