From owner-freebsd-questions Sat Jul 28 22:19: 2 2001 Delivered-To: freebsd-questions@freebsd.org Received: from mail.freebsd-corp-net-guide.com (mail.freebsd-corp-net-guide.com [206.29.169.15]) by hub.freebsd.org (Postfix) with ESMTP id C246537B406 for ; Sat, 28 Jul 2001 22:18:56 -0700 (PDT) (envelope-from tedm@toybox.placo.com) Received: from tedm.placo.com (nat-rtr.freebsd-corp-net-guide.com [206.29.168.154]) by mail.freebsd-corp-net-guide.com (8.11.1/8.11.1) with SMTP id f6T5IV859026; Sat, 28 Jul 2001 22:18:31 -0700 (PDT) (envelope-from tedm@toybox.placo.com) From: "Ted Mittelstaedt" To: "David Powers" , Subject: RE: Network throughput Date: Sat, 28 Jul 2001 22:18:29 -0700 Message-ID: <000501c117ed$e75ffc80$1401a8c0@tedm.placo.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook 8.5, Build 4.71.2173.0 In-Reply-To: <000401c117dd$86a8b860$0401a8c0@daveabit> X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3155.0 Importance: Normal Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Hi, Let me try to answer this from the point of view of the ISP. I'm the admin at the ISP that I work at and we sell DSL. I have my own home DSL connection that goes into the same bridge group that everyone else runs off of. First off, unless the ISP is incredibly oversold, and I mean incredibly, incredibly, oversold there are going to be time periods on the DSL bridge where the average traffic utilization on ALL circuits, both ingress and egress circuits, is going to be down around 5%. With us, that time is about 1:00am to 8:00am. I can pull files off most servers on the Internet at close to 600kps during that time. And we do no caching whatsover. Other times the utilization varies but it is of course much higher. Also bandwidth utilization on the Internet itself is much higher so even if you have a clear path to your ISP and he has a clear path to his backbone feeds, the other end may be stuffed up. So, running these "bandwidth testers" during that time produce basically meaningless results. So what you should learn from this is that IF you are suspecting that your friend's slowness is due to overselling on the ISP, that what you should be seeing then is a tremendous variance on throughput from hour to hour. In short, during the dead times you will see very good throughput and during the hot times it will stink. This should be self-evident, but it's amazing how many users with DSL problems start off blaming the ISP when they get stinky throughput ALL of the time. Now, if your seeing horrible throughput all of the time, then it's not the fault of the ISP and your wasting your time attempting to blame the ISP or it's equipment. Instead, you need to be looking at the DSL circuit itself, AS WELL AS the end user equipment. I've seen some really oddball problems that killed throughout, here they are for your edification: In one customer they had a pair of hubs, one was a switch the other a dumb hub. They complained of piss poor throughput up to complete disconnections. Moving the DSL modem from a switch port to the dumb hub fixed the problem - for all the systems on the dumb hub. The systems on the switch were still having problems. Turned out the switch wasn't designed to handle more than 1 mac address per port and the dumb hub was of course plugged into a port on the switch. (don't ask me why they didn't have problems with their servers) Another customer complained repeatedly over several weeks time. We even rolled a Telco switch tech who tested at the MPOE (demarcation block) and it was fine. Customer kept complaining we finally sent one of our own techs over (gratis, of course) Tech discovered the customer - in an attempt to "neaten up" their server rooom the customer had meticulously braided the DSL modem power cord, telephone line and several other telephone cords together into a 10 foot long bundle. Removed the braiding and separated all cords and throughput went back to normal. In another throughput complaint instance where we ended up sending another tech over gratis (see a pattern here) we found that moving the DSL modem from the top of an Intel hub where it had been resting to a shelf about 3 feet away fixed the problem. One time I got a call from another customer with low throughput complaint. During the course of the call I asked him to describe his inside phone wiring and he admitted he lived in an old house that had been built in 1920 and the phone wire was all original. I told him to run a new CAT-5 cable from the MPOE directly to the DSL modem, then from the DSL modem to the rest of the phones in his house. Fortunately the guy was an EE and readily understood the problems when I explained that many of the DSL frequencies are in the RF range and he complied with instructions. As soon as the new wiring was installed the problems went away. My all-time favorite story was this one, though: Customer called in repeatedly over several weeks time complaining of low throughput, went through the usual rigamarole of sending people out, nothing worked. Finally on the second Telco truck roll I had the tech test from the back of the DSL modem itself instead of the MPOE he discovered a short to another cable pair. Several hours of investigating with a TDR discovered - get this - a HIDDEN intermediate punchdown block concealed behind a false wall in the back of a cabinet in the customer area. They found someone at the site (this was a business) that did the old head-slapper and remembered the story. Seems that 6 years ago they expanded into a second suite and in the process of integrating the two networks ended up with this intermediate punchdown block in an unfavorable location. Said location happened to be the office of an executive secretary (no longer working for them) who was best described as a "bitch on wheels" who got a bug up her butt about "that ugly mess of wires" in the back of one of her cabinets. She hired a carpenter who came in and for $1K or so rebuilt the entire cabinet with the false wall. Idiot carpenter of course provided no door or other means of access to the punchdown block. (they had to use a hammer to smash out the back of this cabinet, fortunately it was crap laminated particle board) Ground short was caused by a nail from the cabinet that had been driven into one of the cable bundles. I think that the phone company charged them about $500 to find that one out. Anyway, what I hope that you draw from all this is that DSL is not a rock-solid circuit delivery mechanism and that the last 30 feet of wire inside the house can kill the circuit as surely as a bridge tap or load coil in the Telco section of the circuit. Ted Mittelstaedt tedm@toybox.placo.com Author of: The FreeBSD Corporate Networker's Guide Book website: http://www.freebsd-corp-net-guide.com >-----Original Message----- >From: owner-freebsd-questions@FreeBSD.ORG >[mailto:owner-freebsd-questions@FreeBSD.ORG]On Behalf Of David Powers >Sent: Saturday, July 28, 2001 8:21 PM >To: freebsd-questions@FreeBSD.ORG >Subject: Network throughput > > >I am trying to figure out the best way to attack this problem. I have a >friend that I'm trying to help out with a DSL connection. Their connection >is obviously slow and I'm looking for some ideas on how I can go about >calculating network throughput to their gateway which is believed to be the >problem. The ISP is giving them quite the run around saying that everything >looks just find from their end. > > >To Unsubscribe: send mail to majordomo@FreeBSD.org >with "unsubscribe freebsd-questions" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message