From owner-freebsd-hackers Fri Dec 8 08:38:20 1995 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id IAA15754 for hackers-outgoing; Fri, 8 Dec 1995 08:38:20 -0800 (PST) Received: from lupine.nsi.nasa.gov (lupine.nsi.nasa.gov [198.116.2.100]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id IAA15746 for ; Fri, 8 Dec 1995 08:38:13 -0800 (PST) Received: (from mnewell@localhost) by lupine.nsi.nasa.gov (8.6.12/8.6.12) id LAA17904; Fri, 8 Dec 1995 11:34:00 -0500 Date: Fri, 8 Dec 1995 11:34:00 -0500 (EST) From: "Michael C. Newell" To: Joe Greco cc: Angelo Turetta , freebsd-hackers@freebsd.org Subject: Re: User space PPP & leased line In-Reply-To: <199512071146.FAA10177@brasil.moneng.mei.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hackers@freebsd.org Precedence: bulk On Thu, 7 Dec 1995, Joe Greco wrote: > > BTW, is there any {performance|machine load|memory usage} gain using one or > > the other of the two ppp daemons ? > > The kernel mode PPP should conceptually be more efficient and robust because > it is built into the kernel. It does not have the same problems of context > switching, using memory per process, etc. etc... > > However, the user mode PPP is easier to debug and hack on. It has some > rather severe bugs for any site that wants to run more than just a few, but > these are known quantities and tend to be easier to fix than kernel bugs. :-) We tried using the user mode PPP on our systems for a while, but under very heavy load after 2-3 minutes the line ALWAYS dropped. No indication of what caused it. We use the identical setup with the kernel PPP (pppd) and run for days with no line drops. We sent mail to the list a couple of times but never got responses, so we just stuck with the kernel mode code. One thing we DID notice was that the transfer speeds using the user mode code were SIGNIFICANTLY higher than those using the kernel mode code - for example, transferring a medium size binary file (one that could be fully transferred in < 2 minutes :-) using pppd takes an average of 2.3KB/s whereas with ijppp it's more like 3.3KB/s (using a V.fc modem connection at 26.4Kb/s). I'm not sure why this should be, but given the improved performance we REALLY wanted ijppp to work. But... ;-( We DID have to set up a bunch of scripts to make sure the kernel mode doesn't lock up and to do redialing ('cause we've never been able to get pppd to redial on its own). They weren't very complicated though so pretty much it wasn't hard to get set up and running. If it would be useful for you I'd be happy to send you copies of the scripts. Thanks, Mike +--------------------------------------+------------------------------------+ |Mike Newell | The opinions expressed herein are | |NASA Science Internet Network Systems | my own, and do not necessarily | |Sterling Software, Inc. | reflect those of the NSI program, | |MNewell@nsipo.nasa.gov | Sterling Software, NASA, or anyone | |+1-202-434-8954 | else. | +--------------------------------------+------------------------------------+ | work: http://www.eco.nsi.nasa.gov/~mnewell | | home: http://www.newell.arlington.va.us | +---------------------------------------------------------------------------+