Return-Path: Received: from mail-oi0-f48.google.com ([209.85.218.48]:34454 "EHLO mail-oi0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753845AbcERVBm (ORCPT ); Wed, 18 May 2016 17:01:42 -0400 Received: by mail-oi0-f48.google.com with SMTP id k142so97738005oib.1 for ; Wed, 18 May 2016 14:01:41 -0700 (PDT) MIME-Version: 1.0 From: Martin Houry Date: Wed, 18 May 2016 23:01:21 +0200 Message-ID: Subject: Question : Session trunking and tload balancing algorithm To: Linux NFS Mailing List Content-Type: text/plain; charset=UTF-8 Sender: linux-nfs-owner@vger.kernel.org List-ID: Hello all, I have a question about session trunking and the algorithm who do load balancing in. Is already some ideas to upgrade the algorithm to fit it with the fluctuation of the bandwith in the network? I have proposed to use the queue's lenght of the RPC call list to get better speed transfert in a previous mail. I have tested it and there is a huge performance difference when the quality of the paths are different. Do you think it's a good idea to add an option to enable this updrade? In other word do you think it's useful for the end-user to choose the algorithm of the session trunking load balancing? I can send you by mail some graph about my test. Thank you Martin