From owner-freebsd-fs@FreeBSD.ORG Tue Sep 18 01:08:12 2012 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: by hub.freebsd.org (Postfix, from userid 821) id 66E1A106566C; Tue, 18 Sep 2012 01:08:12 +0000 (UTC) Date: Tue, 18 Sep 2012 01:08:12 +0000 From: John To: FreeBSD FS Message-ID: <20120918010812.GA71005@FreeBSD.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.4.2.1i Subject: XFS/istgt backed 8TB xfs filesystem configuration? X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 18 Sep 2012 01:08:12 -0000 Hi Folks, I've been asked to export an 8TB volume (via istgt) and was curious if anyone has any experience with optimal xfs configurations in this area. Typically, volumes for linux are created similar to: zfs create -b 32768 -V $lunsize $physname On a dual 10g data backbone network, we use mpio with 2 channels per net: [PortalGroup4] Comment "Two networks - Two ports" Portal DA1 10.59.10.10:5000 Portal DA2 10.60.10.10:5000 Portal DA3 10.59.10.10:5001 Portal DA4 10.60.10.10:5001 Comment "END: PortalGroup4" which typically seems to give the best performance. The luns are being brought together on the linux side (RHEL 6.1) with multipath. I've google'd around a bit and don't see much about zfs filesystems on top of iscsi exported zfs volumes :-) Anyone have any experience in this area? Suggestions? I'm told the data patterns will be mostly database reads, minimal writes. Thanks, John