Date: Mon, 31 Aug 1998 23:42:09 +0000 From: Mike Smith <mike@smith.net.au> To: "Brent Wiese" <brent@vivid.net> Cc: freebsd-smp@FreeBSD.ORG Subject: Re: What would it take? Message-ID: <199808312342.XAA01358@word.smith.net.au> In-Reply-To: Your message of "Mon, 31 Aug 1998 12:12:12 -0400." <008601bdd4fa$1d5d2cc0$c1a1b7ce@home.vivid.net>
next in thread | previous in thread | raw e-mail | index | archive | help
> > I'm now in a situation where I have to put together some extremely powerful > web servers. The site that it is for is in the top 10 of all visited sites > (no, its not a porno site). :) > > Currently, we have 4 single cpu boxes, 2 running linux and 2 running 2.2.6. > The 2 running FreeBSD are dual pII 300's. > > The plan is to get 2 more of the dual cpu boxes and kill off the 2 single > cpu linux boxes. > > What I want to know is, what is it going to take to make the SMP code usable > in a stressed environment? We have a definate interest in this. I know that > FreeBSD using SMP, running apache, could *seriously* outperfom the same box > running NT and IIS (boo hiss!). The first question you have to ask is whether you actually *need* the extra CPU in the first place. Are you CPU bound? Unless you're doing *heavy* CGI stuff, that's quite likely not the case. If it's clear that you really need more CPU muscle, then the next step is to set up a test server using the -current code, and start pounding it. It's not unlikely that the SMP in -current is actually capable of handling your work profile already, but there are also some known problems, and having an actively prodded test system will help locate these. At this point, it may be possible to move forward chasing the bugs that bite your specific application(s) until the system is suitably stable for your uses. This is the "low effort, slow return" path. > Let me know what you'd want from me to make this happen (money, test > computers, etc) and I'll see what I do. Things like the client doing a press > release about how awesome FreeBSD is serving their website, letting you use > their name on your website as a "huge traffic" client, etc are possibilities > too. The key to SMP development is development talent. SMP is complex stuff, and people with the requisite skills are hard to find. You might want to ask around a little outside the strictly FreeBSD community, and see if you can't find someone interested in taking it on as paywork. You'd quite possibly be able to find co-sponsors for such a venture, and I would expect that the Project would be more than happy to work with you on this. -- \\ Sometimes you're ahead, \\ Mike Smith \\ sometimes you're behind. \\ mike@smith.net.au \\ The race is long, and in the \\ msmith@freebsd.org \\ end it's only with yourself. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199808312342.XAA01358>