From mboxrd@z Thu Jan 1 00:00:00 1970 From: Fyodor Ustinov Subject: raid40? Date: Sat, 16 Apr 2011 04:10:52 +0300 Message-ID: <4DA8EC9C.8070605@ufm.su> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from mail.ufm.su ([77.120.103.19]:38247 "EHLO mail.ufm.su" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751289Ab1DPBKz (ORCPT ); Fri, 15 Apr 2011 21:10:55 -0400 Received: from localhost (localhost.localdomain [127.0.0.1]) by mail.ufm.su (Postfix) with ESMTP id DDC0C600307 for ; Sat, 16 Apr 2011 04:10:52 +0300 (EEST) Received: from mail.ufm.su ([127.0.0.1]) by localhost (mail.ufm.su [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id T7evr+y61WGs for ; Sat, 16 Apr 2011 04:10:52 +0300 (EEST) Received: from [192.168.2.22] (unknown [94.232.181.14]) by mail.ufm.su (Postfix) with ESMTPSA id BD7CA6002EA for ; Sat, 16 Apr 2011 04:10:52 +0300 (EEST) Sender: ceph-devel-owner@vger.kernel.org List-ID: To: ceph-devel@vger.kernel.org Hi. Looking the sources I noticed that the "rule" in CRUSH may be type "raid4". Because of this, I have a question: Is it possible to organize the 20 OSD servers in the scheme 4 * (4+1)? I.e. like a stripe from 4 of raid4 sets? Plain "replication level = 2" is too expensive by "used space". WBR, Fyodor.