Date: Wed, 6 Jun 2001 12:32:02 -0500 From: Mike Meyer <mwm@mired.org> To: Calvin NG <calvinng@brel.com> Cc: "Karsten W. Rohrbach" <karsten@rohrbach.de>, freebsd-stable@FreeBSD.ORG Subject: Re: building apache from /usr/ports Message-ID: <15134.26898.890143.63429@guru.mired.org> In-Reply-To: <20010606113119.B54034@brel.com> References: <JBEOKPCEMKJLMJAKBECCMELDDAAA.jwatkins@firstplan.com> <20010605140629.B15206@leviathan.inethouston.net> <20010605152718.A21889@localhost> <20010606034917.D97958@mail.webmonster.de> <15133.37451.23934.758674@guru.mired.org> <20010606113119.B54034@brel.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Calvin NG <calvinng@brel.com> types: > Correct me if I m wrong. > For in-core web server , every copy of server loaded has the perl and/or php > in it. For modules, its a shared library, the server is smaller size, and only > a copy of the module is loaded in memory. That's all correct. However, it wouldn't surprise me if the server + module is larger than the server with an in-core module. Since there's no reason to have more than one copy of the server loaded - even for a high-load server - there's still only one copy of the module loaded. Since everything shared in the module should also be shared with the in-core version, the total memory usage won't be very different. > However, in-core is slightly faster then modules, IIRC. > > I remember reading the performance pages of mod_perl, they recommend running > mod_perl in-core servers separately as a application server. > > Thats my understanding of the difference between in-core and modules. That indicates that there may be some large data structure that's initialized per process. The only reason it should use a lot more memory in-core than as a module is if this happens when the module is loaded, and not when it is enabled. <mike > Regards, > /calvin > > lines with :> are quotes from Mike Meyer's email > :> Karsten W. Rohrbach <karsten@rohrbach.de> types: > :> > you won't recognize it until you have to implement a heavily loaded > :> > server with php or perl in-core. position independent code is know to be > :> > slower, but it outperforms monolithic compiles by saving a lot of ram. > :> > :> Ok, where does the savings come from? You get one copy of the code, > :> shared by all the processes running the binary, whether or not the > :> code is in a shared library. COW for data should mean that data should > :> be shared pretty much the same. So what have I missed? > :> > :> Thanx, > :> <mike > :> -- > -- Mike Meyer <mwm@mired.org> http://www.mired.org/home/mwm/ Independent WWW/Perforce/FreeBSD/Unix consultant, email for more information. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15134.26898.890143.63429>