From owner-freebsd-virtualization@FreeBSD.ORG Sun Feb 2 06:19:16 2014 Return-Path: Delivered-To: freebsd-virtualization@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 65107F2D for ; Sun, 2 Feb 2014 06:19:16 +0000 (UTC) Received: from alto.onthenet.com.au (alto.OntheNet.com.au [203.13.68.12]) by mx1.freebsd.org (Postfix) with ESMTP id 2226710CD for ; Sun, 2 Feb 2014 06:19:15 +0000 (UTC) Received: from dommail.onthenet.com.au (dommail.OntheNet.com.au [203.13.70.57]) by alto.onthenet.com.au (Postfix) with ESMTPS id 3350012268; Sun, 2 Feb 2014 16:19:08 +1000 (EST) Received: from Peter-Grehans-MacBook-Pro-2.local (c-67-161-27-37.hsd1.ca.comcast.net [67.161.27.37]) by dommail.onthenet.com.au (MOS 4.2.4-GA) with ESMTP id BRO19515 (AUTH peterg@ptree32.com.au); Sun, 2 Feb 2014 16:19:06 +1000 Message-ID: <52EDE352.9050805@freebsd.org> Date: Sat, 01 Feb 2014 22:18:58 -0800 From: Peter Grehan User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:24.0) Gecko/20100101 Thunderbird/24.2.0 MIME-Version: 1.0 To: Aryeh Friedman Subject: Re: some interesting observations on the relative performance of kvm vs. bhyve References: <52EDB899.9060703@freebsd.org> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: "freebsd-virtualization@freebsd.org" X-BeenThere: freebsd-virtualization@freebsd.org X-Mailman-Version: 2.1.17 Precedence: list List-Id: "Discussion of various virtualization techniques FreeBSD supports." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 02 Feb 2014 06:19:16 -0000 Hi Aryeh, > Does this bring up the same power failure scenario issues mentioned in > the link you provided? It seems like the only way to get reasonable > performance is to be essentially unsafe in guest writes to the host > disk? Use ZFS and a ZIL to mitigate this. Or UFS with journalling. > A question does the ability of FreeBSD to be able to better > handle power failure in general better then linux (it seems like every > time there is a unscheduled reboot on linux it messes up)? It's not the o/s but the behavior of the filesystem. > This seems > to be at odds with my personal observations of bhyve via petitecloud > which I routinely very abruptly start/stop (petitecloud's "stop" is > nothing more then killing the hyperv and any cleanup needed) and except > for the occasional need for a fsck have not had an issue. It's not the VM stop/crash that's the problem, but the host. Once a write has been marked as complete by the VM, the guest assumes it has been committed to stable storage and can assume e.g. that a database transaction is complete. If the write is being held in the host's buffer cache and power is lost, the write never makes it to stable storage, resulting in data loss/corruption for the guest. This is different than the VM crashing, since the writes in the host's buffer cache aren't lost in that case. later, Peter.