From mboxrd@z Thu Jan 1 00:00:00 1970 From: Josh Durgin Subject: Re: use striping feature Date: Mon, 05 Nov 2012 11:38:49 -0800 Message-ID: <509815C9.5090106@inktank.com> References: <5097AAF7.8080709@profihost.ag> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from mail-pb0-f46.google.com ([209.85.160.46]:39142 "EHLO mail-pb0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751180Ab2KETjJ (ORCPT ); Mon, 5 Nov 2012 14:39:09 -0500 Received: by mail-pb0-f46.google.com with SMTP id rr4so4150640pbb.19 for ; Mon, 05 Nov 2012 11:39:09 -0800 (PST) In-Reply-To: <5097AAF7.8080709@profihost.ag> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: Stefan Priebe - Profihost AG Cc: "ceph-devel@vger.kernel.org" On 11/05/2012 04:03 AM, Stefan Priebe - Profihost AG wrote: > Hello list, > > is the following syntax to use the new rbd v2 striping feature correct? > > rbd create -p kvmstor --format 2 --size 32000 --stripe-count 8 > --stripe-unit 131072 -s 1048576 $imagename > > Idea is to have 8 stripes per 1MB object size. Or are there any other > recommondations regarding striping? The stripe count is actually how many objects to stripe over before repeating back to the first one. It's the number of objects in an 'object set' in the diagram at the bottom of: http://ceph.com/docs/master/dev/file-striping/ I don't think we've done enough benchmarks to know what the best settings are, but generally having large (like 4MB) objects and smaller stripe units will at least help small sequential I/O. > I'm a bit confused cause rbd info shows stripe-count as bytes? It's not actually bytes. Sage just fixed that output in the master branch. Josh