From owner-freebsd-fs@FreeBSD.ORG Fri Feb 15 11:01:30 2013 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.FreeBSD.org [8.8.178.115]) by hub.freebsd.org (Postfix) with ESMTP id 7517E61C for ; Fri, 15 Feb 2013 11:01:30 +0000 (UTC) (envelope-from daniel@digsys.bg) Received: from smtp-sofia.digsys.bg (smtp-sofia.digsys.bg [193.68.3.230]) by mx1.freebsd.org (Postfix) with ESMTP id D6FA9189 for ; Fri, 15 Feb 2013 11:01:29 +0000 (UTC) Received: from [10.5.0.53] ([87.250.40.22]) (authenticated bits=0) by smtp-sofia.digsys.bg (8.14.5/8.14.5) with ESMTP id r1FB1NM1036329 (version=TLSv1/SSLv3 cipher=AES128-SHA bits=128 verify=NO); Fri, 15 Feb 2013 13:01:23 +0200 (EET) (envelope-from daniel@digsys.bg) Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 (Mac OS X Mail 6.2 \(1499\)) Subject: Re: Crazy ZFS ZIL options: md(4) umass(4) From: Daniel Kalchev In-Reply-To: Date: Fri, 15 Feb 2013 12:01:23 +0100 Content-Transfer-Encoding: quoted-printable Message-Id: References: To: grarpamp X-Mailer: Apple Mail (2.1499) Cc: freebsd-fs@freebsd.org X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 15 Feb 2013 11:01:30 -0000 Some further clarification. ZIL is only read on crash. If you shut down properly etc, the ZIL will = be never read. For most of it's lifetime, the ZIL is write-only safety = net. An insurance, if you wish. Because it is write-only, cheap USB FLASH devices are the poorest = candidates. USB FLASH devices, even the most "high performance" are = mediocre at best. Because SLOG (separate ZIL) is written sequentially, = even the slowest spinning disk has higher throughput. You want = "write-optimised" device for the SLOG, which no USB FLASH device is, = ever. By the way, you may have better luck with professional photography = write-optimised CF cards, that are capable of recording large amounts of = huge raw images in a burst -- but these typically are more expensive = than an write-optimised SSD and you don't really need much capacity. But = again: forget about USB if you don't enjoy problems. :) By separating the ZIL, you avoid fragmentation, yes. This is because = in-pool ZIL uses variable length records, that are then removed, leaving = variable length holes. ZFS tries to work around this, but it's never = perfect. If you use separate mirrored disks for the ZIL, you will=20 - keep writing sequential (no seeks from interfering reads); - have a reliable recovery medium for the ZIL as long as at least one of = the drives survives the outage. If you want ultimate performance, the solution is battery backed RAM = drives for ZIL. Daniel