From owner-freebsd-net@FreeBSD.ORG Thu Oct 12 07:30:04 2006 Return-Path: X-Original-To: freebsd-net@freebsd.org Delivered-To: freebsd-net@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id CA35316A4A7 for ; Thu, 12 Oct 2006 07:30:04 +0000 (UTC) (envelope-from ivsan@ngs.ru) Received: from intranet.ru (intranet.ru [212.164.71.24]) by mx1.FreeBSD.org (Postfix) with ESMTP id B660B43D72 for ; Thu, 12 Oct 2006 07:29:52 +0000 (GMT) (envelope-from ivsan@ngs.ru) Received: from [172.16.1.1] (HELO mx1.intranet.ru) by intranet.ru (CommuniGate Pro SMTP 4.3.2) with ESMTP id 308739804 for freebsd-net@freebsd.org; Thu, 12 Oct 2006 14:29:49 +0700 Received: from [80.242.64.3] (account ivsan@ngs.ru) by mx1.intranet.ru (CommuniGate Pro WebUser 4.3.2) with HTTP id 80365442 for freebsd-net@freebsd.org; Thu, 12 Oct 2006 14:29:49 +0700 From: "Ivan Alexandrovich" To: freebsd-net@freebsd.org X-Mailer: CommuniGate Pro WebUser Interface v.4.3.2 Date: Thu, 12 Oct 2006 14:29:49 +0700 Message-ID: In-Reply-To: <452C268E.4020700@mavhome.dp.ua> References: <1160493809.00616691.1160482801@10.7.7.3> <452C268E.4020700@mavhome.dp.ua> MIME-Version: 1.0 Content-Type: text/plain; charset="KOI8-R"; format="flowed" Content-Transfer-Encoding: 8bit Subject: Re: ng_netflow and router performance question X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 12 Oct 2006 07:30:04 -0000 On Wed, 11 Oct 2006 02:02:38 +0300 Alexander Motin wrote: > I think, that there is not very good hash function now used in ng_netflow >in traffic aggregation. So if > > ip-addr varies from 10.60.0.0 to 10.60.100.255 > means than destination address will vary in this range and all other >parameters is remain constant then it will be worst case possible. Thanks for your help. With pretty random src ip (10.0.*.* - 100.*.*.*) it was able to handle 23K pkt/s of unique flows without packet losses and with 99,96 accuracy (both active and inactive timeouts were set to 3 seconds for testing purposes). I'd like to ask about the reasonable values of timeout parameters for a highly loaded router to avoid records cache overruns? There is a compile-time option CACHESIZE defined in ng_netflow.h. Is it ok to increase it or should I manipulate with timeout values alone? Thanks, Ivan