From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andi Kleen Subject: Re: [ofa-general] [PATCH 0/26] Reliable Datagram Sockets (RDS), take 2 Date: Sat, 28 Feb 2009 23:36:53 +0100 Message-ID: <20090228223653.GD26292@one.firstfloor.org> References: <1235525443-9007-1-git-send-email-andy.grover@oracle.com> <87myc73izx.fsf@basil.nowhere.org> <20090228055608.GB26292@one.firstfloor.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Andi Kleen , Andy Grover , netdev@vger.kernel.org, rds-devel@oss.oracle.com, general@lists.openfabrics.org To: Andrew Grover Return-path: Received: from one.firstfloor.org ([213.235.205.2]:56618 "EHLO one.firstfloor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754526AbZB1WSY (ORCPT ); Sat, 28 Feb 2009 17:18:24 -0500 Content-Disposition: inline In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: > The previous solution for IPC that Oracle was using was based on UDP, > which I think could be considered very close to using raw sockets -- > each process is responsible for its own acks, retransmits, everything. > Doing this on a highly loaded machine resulted in a cascade where > performance got worse and worse. Could you describe that cascade in more detail? The problem was that the retransmits didn't have high enough priority? > Additionally, our interconnect is primarily Infiniband. It natively > implements a reliable datagram connection type so RDS leverages that. So perhaps it would make more sense to have a thin direct interface to that IB service? Or perhaps it already exists? (I admit I don't know the IB interfaces very well) -andi -- ak@linux.intel.com -- Speaking for myself only.