Date: Sun, 10 Dec 2006 09:19:59 -0600 From: Lane <lane@joeandlane.com> To: freebsd-questions@freebsd.org Subject: Re: Advantages of trimmed kernel? Message-ID: <200612100919.59564.lane@joeandlane.com> In-Reply-To: <200612100905.30430.kirk@strauser.com> References: <200612100905.30430.kirk@strauser.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sunday 10 December 2006 09:05, Kirk Strauser wrote: > Are there any real advantages to building a kernel stripped of unused > drivers, especially when running it on a fairly large machine? For years, > I've been dutifully removing device drivers (or more recently, including > GENERIC and using 'nodevice') for everything I don't have. But does this > actually do anything useful, or am I just tilting at windmills? > > I know the definitive answer would be to run benchmarks both ways, but I > don't really have the option of pulling down a production machine just for > this. Kirk, I don't expect there is only one answer to your question. The issue is broader, I think, than just the relative speed and performance improvements achieved by running a lean kernel. You say that you can't afford to take a production machine down, but consider this: What if you trimmed all of the "fat" from the kernel on a server, and then the server's nic goes bad. Suppose that as a stop-gap measure you pull an old isa nic from out of the closet, install it, and then boot the server ... only to realize that your nic is not supported by the kernel that you dutifully trimmed. I think it is especially important to keep the kernel as flexible as possible, since you may have to install the OS on any given machine without the luxury of recompiling. Just my .02 lane
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200612100919.59564.lane>