From owner-freebsd-questions@FreeBSD.ORG Fri Feb 8 17:22:43 2008 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 9EFB516A418 for ; Fri, 8 Feb 2008 17:22:43 +0000 (UTC) (envelope-from wundram@beenic.net) Received: from mail.beenic.net (mail.beenic.net [83.246.72.40]) by mx1.freebsd.org (Postfix) with ESMTP id 5DB7113C461 for ; Fri, 8 Feb 2008 17:22:43 +0000 (UTC) (envelope-from wundram@beenic.net) Received: from phoenix (hnvr-4dbb95fb.pool.einsundeins.de [77.187.149.251]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mail.beenic.net (Postfix) with ESMTP id 96D89A44529 for ; Fri, 8 Feb 2008 18:22:34 +0100 (CET) From: "Heiko Wundram (Beenic)" Organization: Beenic Networks GmbH To: freebsd-questions@freebsd.org Date: Fri, 8 Feb 2008 18:24:03 +0100 User-Agent: KMail/1.9.7 References: <484700.62092.qm@web34508.mail.mud.yahoo.com> In-Reply-To: <484700.62092.qm@web34508.mail.mud.yahoo.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200802081824.03290.wundram@beenic.net> Subject: Re: Some ideas for FreeBSD X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 08 Feb 2008 17:22:43 -0000 Am Freitag, 8. Februar 2008 17:54:03 schrieb millueradfa@yahoo.com: > Well, actually, these are file backed swap devices. > You can do both file and memory backed devices. this > allows you to have a swap file on the hard disk and > mount it. As I already wrote in another part of this thread: please explain to me why it should be faster to have a file backed md set up as swap than a dedicated swap partition (because there's at least two more levels of indirection involved). I can clearly see the need for file backed swap in special cases (for example, where you need RAM desperately, for example for a compile, but cannot add another partition to a system), but no matter what, it will never be faster than a swap partition. And that was what the original poster of this sub-thread suggested (and as such, I took it that he was referring to memory-backed mds, because file-backed mds are never faster than "raw" access to a hard-disk). So, I still stand by my first assessment: the idea to use an md as swap is stupid, at least from a performance standpoint. -- Heiko Wundram Product & Application Development