From owner-freebsd-geom@FreeBSD.ORG Wed Mar 5 18:12:49 2014 Return-Path: Delivered-To: freebsd-geom@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 73780E38; Wed, 5 Mar 2014 18:12:49 +0000 (UTC) Received: from mail.tdx.com (mail.tdx.com [62.13.128.18]) by mx1.freebsd.org (Postfix) with ESMTP id 14F51AD5; Wed, 5 Mar 2014 18:12:48 +0000 (UTC) Received: from study64.tdx.co.uk (study64.tdx.co.uk [62.13.130.231]) (authenticated bits=0) by mail.tdx.com (8.14.3/8.14.3/) with ESMTP id s25ICkOs052545 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Wed, 5 Mar 2014 18:12:47 GMT Date: Wed, 05 Mar 2014 18:12:46 +0000 From: Karl Pielorz To: Mark Felder , freebsd-geom@freebsd.org Subject: Re: HAST local read performance? Message-ID: <0256F5C22685BD8205AE3952@study64.tdx.co.uk> In-Reply-To: <1394035980.8640.90922737.615AFA56@webmail.messagingengine.com> References: <1394035980.8640.90922737.615AFA56@webmail.messagingengine.com> X-Mailer: Mulberry/4.0.8 (Mac OS X) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline X-BeenThere: freebsd-geom@freebsd.org X-Mailman-Version: 2.1.17 Precedence: list List-Id: GEOM-specific discussions and implementations List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 05 Mar 2014 18:12:49 -0000 --On 5 March 2014 10:13:00 -0600 Mark Felder wrote: > Which HAST replication mode are you using? Fullsync? Do you have atime > enabled on that filesystem? I'd expect it to act like a local disk > without any significant penalties unless you're doing writes and they're > waiting to be synced before the next read. Currently using async (makes no difference from fullsync - I'd guess that would only affect writes, not reads), and 'checksum none' No atime enabled on the file system (ZFS) - the speed difference is very apparent with reading via dd (which is obviously raw-read only) - and on zpool scrubbing (which is 99% read with few writes) - if I deliberately 'fail' the secondary (so not even those few writes are sent to the secondary) the speed remains the same. It's certainly 'usable' - the only issue is with a few TB's of ZFS filesystem to scrub, it's rather slow through HAST (even when reads are happening only on the local machine). Like you said you'd expect reads, from the local primary would be pretty similar to a local disk. I guess with more devices online it'll pickup speed wise anyway. -Karl