From owner-freebsd-atm Tue Nov 16 18:20: 8 1999 Delivered-To: freebsd-atm@freebsd.org Received: from piglet.dstc.edu.au (piglet.dstc.edu.au [130.102.176.1]) by hub.freebsd.org (Postfix) with ESMTP id 6487914E20 for ; Tue, 16 Nov 1999 18:19:57 -0800 (PST) (envelope-from ggm@dstc.edu.au) Received: from azure.dstc.edu.au (azure.dstc.edu.au [130.102.176.27]) by piglet.dstc.edu.au (8.9.3/8.9.3) with ESMTP id MAA12784 for ; Wed, 17 Nov 1999 12:19:56 +1000 (EST) Received: (from ggm@localhost) by azure.dstc.edu.au (8.8.5/8.8.4) id MAA19896 for freebsd-atm@freebsd.org; Wed, 17 Nov 1999 12:19:56 +1000 (EST) Date: Wed, 17 Nov 1999 12:19:56 +1000 (EST) From: George Michaelson Message-Id: <199911170219.MAA19896@azure.dstc.edu.au> To: freebsd-atm@freebsd.org Subject: Bandwidth Capping fore ATM circuits? Sender: owner-freebsd-atm@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org I have a 2-PVC network config between two sites using Fore PCA200e cards. Well, its one PVC between the sites, and one additional PVC at the "main" site to provide edge connect into an Internet cloud. My backend network provider (campus networking) reports that I am exceeding their bandwidth allocations for the inter-site PVC, and the behaviour I see is that FTP downloads stall badly (presumably because TCP sees horrendous drop rate) I can't see any code clues to how I can do capped bandwidth stuff on this setup. Help! I need to make a 155mbps edge device behave "well" on 24mbps core network bandwidth... cheers -George To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-atm" in the body of the message From owner-freebsd-atm Wed Nov 17 1:20:49 1999 Delivered-To: freebsd-atm@freebsd.org Received: from pedigree.cs.ubc.ca (pedigree.cs.ubc.ca [142.103.6.50]) by hub.freebsd.org (Postfix) with ESMTP id 0C525152BE for ; Wed, 17 Nov 1999 01:20:43 -0800 (PST) (envelope-from mjmccut@cs.ubc.ca) Received: (from ean@localhost) by pedigree.cs.ubc.ca (8.8.8/8.6.9) id BAA24856 for freebsd-atm@freebsd.org; Wed, 17 Nov 1999 01:20:36 -0800 (PST) X400-Received: by /PRMD=ca/ADMD=telecom.canada/C=ca/; Relayed; Wed, 17 Nov 1999 1:20:35 UTC-0800 Date: Wed, 17 Nov 1999 1:20:35 UTC-0800 X400-Originator: mjmccut@cs.ubc.ca X400-Recipients: non-disclosure:; X400-Content-Type: P2-1984 (2) X400-MTS-Identifier: [/PRMD=ca/ADMD=telecom.canada/C=ca/;991117012035] Content-Identifier: 16609 From: Mark McCutcheon To: freebsd-atm@freebsd.org Message-ID: <"16609*mjmccut@cs.ubc.ca"@MHS> Subject: Re: Bandwidth Capping fore ATM circuits? MIME-Version: 1.0 (Generated by Ean X.400 to MIME gateway) Sender: owner-freebsd-atm@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Wed, 17 Nov 1999 12:19:56 UTC+1000 George Michaelson wrote: > I have a 2-PVC network config between two sites > using Fore PCA200e cards. Well, its one PVC between > the sites, and one additional PVC at the "main" site > to provide edge connect into an Internet cloud. > > My backend network provider (campus networking) reports > that I am exceeding their bandwidth allocations for the > inter-site PVC, and the behaviour I see is that FTP downloads > stall badly (presumably because TCP sees horrendous drop rate) > > I can't see any code clues to how I can do capped bandwidth stuff > on this setup. > > Help! I need to make a 155mbps edge device behave "well" on 24mbps > core network bandwidth... I imagine you're using the HARP ATM networking code supplied with current versions of FreeBSD. I haven't used it but understand that it does not provide native traffic control facilities. If you need to do native ATM bandwidth management, that's a problem. If you can use IP-level bandwidth control, 2 means suggest themselves: a) Luigi Rizzo's dummynet pipes under ipfw (man pages in current versions) - haven't used it but there are favourable reports. See also http://www.iet.unipi.it/~luigi/ip_dummynet/ b) Kenjiro Cho's Alternate Queuing (ALTQ) with Class-Based Queuing (CBQ) allows close control of bandwidth in IP flows. Have used this and it works well with 100BaseT. However, it needs to modify the output NIC drivers, and in earlier versions only knew about Adaptec/Efficient ATM NICs with Chuck Cranor's BSD ATM. You could see whether the current release has support for Fore NICs and HARP ATM code.... http://www.csl.sony.co.jp/person/kjc/programs.html Hope this helps, Mark To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-atm" in the body of the message From owner-freebsd-atm Wed Nov 17 1:38:28 1999 Delivered-To: freebsd-atm@freebsd.org Received: from praam.cs.vu.nl (praam.cs.vu.nl [192.31.231.78]) by hub.freebsd.org (Postfix) with ESMTP id 8CD3D14D02 for ; Wed, 17 Nov 1999 01:38:24 -0800 (PST) (envelope-from versto@cs.vu.nl) Received: by praam.cs.vu.nl (Smail #62) id m11o1X7-000HmVC; Wed, 17 Nov 99 10:37 +0100 From: "Kees Verstoep" Message-Id: <9911171037.ZM24889@cs.vu.nl> Date: Wed, 17 Nov 1999 10:37:45 +0100 In-Reply-To: Mark McCutcheon "Re: Bandwidth Capping fore ATM circuits?" (Nov 17, 1:20am) References: <"16609*mjmccut@cs.ubc.ca"@MHS> X-Mailer: Z-Mail (3.2.1 10apr95) To: freebsd-atm@freebsd.org Subject: Re: Bandwidth Capping fore ATM circuits? Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-freebsd-atm@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Wed, 17 Nov 1999 12:19:56 UTC+1000 George Michaelson wrote: >[..] > I can't see any code clues to how I can do capped bandwidth stuff > on this setup. > > Help! I need to make a 155mbps edge device behave "well" on 24mbps > core network bandwidth... In general it should be quite easy to modify any driver for the Fore 200E to enforce some kind of rate limiting. It has native support for limiting the per-PDU outgoing cell rate by means of a simple "data/idle cell" specification. If you don't need a flexible API that allows you to change this rate on the fly, it'll probably just take a two-line modification to set it to the rate you want. For an example how this is done you can look at the Fore 200E driver for Linux, available from http://cedric.cnam.fr/personne/lizzi/linux/fore200e.html Good luck, Kees Verstoep To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-atm" in the body of the message From owner-freebsd-atm Thu Nov 18 15:39:48 1999 Delivered-To: freebsd-atm@freebsd.org Received: from piglet.dstc.edu.au (piglet.dstc.edu.au [130.102.176.1]) by hub.freebsd.org (Postfix) with ESMTP id CC8F01553D for ; Thu, 18 Nov 1999 15:39:43 -0800 (PST) (envelope-from ggm@dstc.edu.au) Received: from azure.dstc.edu.au (azure.dstc.edu.au [130.102.176.27]) by piglet.dstc.edu.au (8.9.3/8.9.3) with ESMTP id JAA21727 for ; Fri, 19 Nov 1999 09:39:42 +1000 (EST) Received: (from ggm@localhost) by azure.dstc.edu.au (8.8.5/8.8.4) id JAA22574 for freebsd-atm@freebsd.org; Fri, 19 Nov 1999 09:39:40 +1000 (EST) Date: Fri, 19 Nov 1999 09:39:40 +1000 (EST) From: George Michaelson Message-Id: <199911182339.JAA22574@azure.dstc.edu.au> To: freebsd-atm@freebsd.org Subject: Interop problems with Centillion Sender: owner-freebsd-atm@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org I've been told that in some situations the fore NIC is known to have interop problems with Centillion ATM switches. This can be fixed by using later revs microcode. I have been recommended to use 4.0.3.03 or later. I have this nasty feeling that I am using a 4.0.2 release and that using later ones is not expected to work. Can somebody tell me if there is a known-good release which is later than 4.0.3, and which I can use? Arg! cheers -George To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-atm" in the body of the message From owner-freebsd-atm Thu Nov 18 20:21:42 1999 Delivered-To: freebsd-atm@freebsd.org Received: from piglet.dstc.edu.au (piglet.dstc.edu.au [130.102.176.1]) by hub.freebsd.org (Postfix) with ESMTP id 66213154D4 for ; Thu, 18 Nov 1999 20:21:35 -0800 (PST) (envelope-from ggm@dstc.edu.au) Received: from dstc.edu.au (asuncion.dstc.edu.au [130.102.176.155]) by piglet.dstc.edu.au (8.9.3/8.9.3) with ESMTP id OAA21448; Fri, 19 Nov 1999 14:21:26 +1000 (EST) To: "Kees Verstoep" Cc: freebsd-atm@FreeBSD.ORG Subject: Re: Bandwidth Capping fore ATM circuits? In-reply-to: Your message of "Wed, 17 Nov 1999 10:37:45 +0100." <9911171037.ZM24889@cs.vu.nl> Date: Fri, 19 Nov 1999 14:21:25 +1000 Message-ID: <6046.942985285@dstc.edu.au> From: George Michaelson Sender: owner-freebsd-atm@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org This is what I think you said :-) /sys/dev/hfa/fore_output.c: /* * don't do this... */ #ifdef GGM xdp->xd_rate = FORE_DEF_RATE; #else /* * given target max_pcr of 74460 I have to work out * the rate, where rate is the fusion of data and idle as 16-bit fields * joined in a 32. * * Linux does this: * *fore200e_rate_ctrl(struct atm_qos* qos, struct tpd_rate* rate) *{ * if (qos->txtp.max_pcr < ATM_OC3_PCR) { * * compute the data cells to idle cells ratio from the PCR * rate->data_cells = * qos->txtp.max_pcr * FORE200E_MAX_BACK2BACK_CELLS / * ATM_OC3_PCR; * rate->idle_cells = FORE200E_MAX_BACK2BACK_CELLS - * rate->data_cells; * } * else { * disable rate control * rate->data_cells = rate->idle_cells = 0; * } *} * * typedef struct tpd_rate * { * BITFIELD2( * u32 idle_cells : 16, number of idle cells to insert * u32 data_cells : 16 number of data cells to transmit * ) * } tpd_rate_t; * * values found for the #defines from Linux are: * * #define FORE200E_MAX_BACK2BACK_CELLS 255 * * and from atm_h for FreeBSD I get: * * #define ATM_PCR_OC3C 353207 Peak Cell Rate for OC3c * * so I used BC to calculate the date_cells and idle_cells values * as (74460*255/353207)=53 and (255-53)=202 respectively. * on a PC, this becomes a u_long value of 3473610 */ xdp->xd_rate = 3473610; #endif -- George Michaelson | DSTC Pty Ltd Email: ggm@dstc.edu.au | University of Qld 4072 Phone: +61 7 3365 4310 | Australia Fax: +61 7 3365 4311 | http://www.dstc.edu.au To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-atm" in the body of the message