From mboxrd@z Thu Jan 1 00:00:00 1970 From: Loic Dachary Subject: Re: Question on Ceph LRC design Date: Mon, 17 Nov 2014 02:04:00 +0100 Message-ID: <54694980.2010302@dachary.org> References: <06681238D8946F44A60AA400760A1CBF01E28FBD@SHSMSX104.ccr.corp.intel.com> <06681238D8946F44A60AA400760A1CBF01E290AB@SHSMSX104.ccr.corp.intel.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="uSGWXxPxrAGEvNNgJLbRXopv68cLknnIF" Return-path: Received: from mail2.dachary.org ([91.121.57.175]:37910 "EHLO smtp.dmail.dachary.org" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751734AbaKQBED (ORCPT ); Sun, 16 Nov 2014 20:04:03 -0500 In-Reply-To: <06681238D8946F44A60AA400760A1CBF01E290AB@SHSMSX104.ccr.corp.intel.com> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: "Zhou, Yuan" , "Andreas.Joachim.Peters@cern.ch" Cc: "ceph-devel@vger.kernel.org" This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --uSGWXxPxrAGEvNNgJLbRXopv68cLknnIF Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Hi, I believe Andreas has a more elaborate answer on this topic. The current = implementation is not as good as what is described in the paper you menti= on, this is correct. My incentive for chosing this path is that I was abl= e to understand it, mainly. It is not much more than stacking layers of e= rasure coded chunks on top of each other ;-) Now that we have this plugin, it would be nice to have another implementa= tion that uses less space and possible even less network when reconstruct= ing. During the OpenStack summit we discussed this with Kevin Greenan and= there are promising directions. It would help a lot to have a sample cod= e to study so that it can be adapted to what we have currently in Ceph. D= o you know of such an implementation of LRC or other similar code designe= d to reduce the network bandwidth during reconstruction ? Cheers On 17/11/2014 01:52, Zhou, Yuan wrote: > Hi Loic/Anderas, >=20 > =20 >=20 > I was trying to understand the LRC design in Ceph EC. Per my understand= ing, it seems Ceph was using a slightly different design with the Microso= ft LRC: the local parities were calculated with the global parities inclu= ded. Is there any special consideration on this change? >=20 > I was asking because in a typical MS LRC design the global and local pa= rities could be calculated at the same time actually(I mean inside the Er= asure Code library). But with this new design, we lost this potential opt= imization. >=20 > =20 >=20 > Thanks, -Yuan >=20 --=20 Lo=EFc Dachary, Artisan Logiciel Libre --uSGWXxPxrAGEvNNgJLbRXopv68cLknnIF Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlRpSYAACgkQ8dLMyEl6F20uEgCeIquoXQRBCHG3cKkXaWsZNAeu 1SYAn1m+Jt85Gw9flV7Bcx8c8UBlyPij =4lYI -----END PGP SIGNATURE----- --uSGWXxPxrAGEvNNgJLbRXopv68cLknnIF--