From mboxrd@z Thu Jan 1 00:00:00 1970 From: David L Stevens Subject: Re: [PATCHv6 net-next 1/3] sunvnet: upgrade to VIO protocol version 1.6 Date: Mon, 22 Sep 2014 00:40:27 -0400 Message-ID: <541FA83B.2050608@oracle.com> References: <541A2316.5010603@oracle.com> <2AB76E42-C12D-47C5-8476-0D0C611691A5@oracle.com> <541AD838.50700@oracle.com> <9BA1705F-0C89-471A-9872-688A3FA3165C@oracle.com> <541B395D.1000809@oracle.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: David Miller , netdev@vger.kernel.org To: Raghuram Kothakota Return-path: Received: from userp1040.oracle.com ([156.151.31.81]:21590 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750723AbaIVEkd (ORCPT ); Mon, 22 Sep 2014 00:40:33 -0400 In-Reply-To: <541B395D.1000809@oracle.com> Sender: netdev-owner@vger.kernel.org List-ID: > On 09/18/2014 02:49 PM, Raghuram Kothakota wrote: >> In the virtualization world, we want resources to be efficiently used and memory is >> still very important resource. My concern is mostly because this memory usage of >> 32+MB is on a per LDC basis. LDoms today supports a max of 128 domains, but >> from my experience seen actual deployments of the order of 50 domains. This is >> going up as the platforms getting more and more powerful. If there are really >> that many peers, then the amount of memory consumed by one vnet instance >> is 50 * 32+MB = 1.6GB+. It's fine if this memory is really used, but it seems like this >> will be useful only when the peer is another linux guest with this version of vnet and >> also the MTU is configured to use 64K. The memory is being wasted for all other >> peers that either don't support 64K MTU or not configured to use it and also >> the switch port as obviously it doesn't support 64K MTU today. I think I have a solution for this -- I'm doing some experimenting, but it may be a few days. However, fundamentally, the problem is that there are n^2-n links both ways, so 50 LDOMs on the same vswitch will always be 2450X the resources of a single pair, and lead to scary aggregate numbers. Large installations really need more vswitches with fewer LDOMs per switch, at least with the current code. +-DLS