In recent weeks I frequently have periods where all of the ping times from the first node on go over 5000ms. I've discussed this at length with both my ISP and the phone company. The ISP says the delay is between my computer and their logon node. The phone company says that there is nothing wrong with their line and that the delay is at the ISP's initial node. This condition occurs periodically on two different phone lines using either of two different laptops. It occurs even when I unplug the house wiring entirely and plug directly into the phome company test socket in their line termination box.<br><br>This problem is serious and I need some guidance on isolating responsibility for the problem.<br><br>

Hi.<br><br>5000 ms ping times are pretty common on a dial-up when under load (ie: you're saturating the badwidth). With a regular dial-up line, you'll see high ping times *OFTEN* due to this being your bottleneck. During these periods of problems, do you have any other applications transferring data? <br><br>If your computer isn't doing anything else, the troubleshooting between you and hop 1 (the modem line) is a tough one to troubleshoot as your hardware plays into the mix quite heavily. It could be a configuration problem on your end, or it could be a configuration problem from the ISP - or anything in between. I'd probably start out by connecting to a different ISP to see if the connection quality is similar. One of the free services like Net zero should work well for this - or use one of the free offers from AOL, compuserve, or another similar network.<br><br>With ping times of 5000, it's likely it's not the physical phone line. If the quality was that bad, you'd be getting completely dropped connections. If there's no possibility that you're saturating your bandwidth yourself, then my next efforts would be spent on determining if it's a problem with your local hardware or configuration (on your laptop(s)).<br><br>You didn't offer a lot of information on the problem (no performance graphs, no stories about connection problems, or how often the performance is acceptable). More information could help diagnose this better.<br><br>

I just installed the new 2.2 with the graph across the bottom. I have it recording the initial node only and the hourly graph clearly illustrates the problem - am emailing it since I can't post it here. What it shows is occasional long bursts of poor times.<br><br>The connection is under continuous load from a highly compressed quote stream rather than burst loading. I have seen this same behavior when I have no load at all. I have in fact, frequently had connections deteriorate from >5000 times to full ERR and then drop. Unfortunately, I am in a remote area where we have only one ISP available so I don't have alternatives.<br><br>This graph is coming from a new Dell I7500 with Lucent Technologies WinModem (internal) connected at 50200. OS is W2000 professional. Not sure what information would provide clues regarding a configuration problem.<br><br>

Here's the graph sent...<br><br><br><br>First off, the following suggestions are what I would do if I was having the problems you are. These may not be the best or only way of doing this, but ... it's just what I would do.<br><br>Honestly, these performance peaks look *EXACTLY* like modem bandwidth saturation. I've seen this on my own systems many a time, but can always attribute it to something running on my system using bandwidth.<br><br>The very first thing I would do, is to try to find a piece of software that can measure your bandwidth useage on your modem. I don't have a recommendation for a specific one, but I've seen many that basically replace the little tool-tray lights for the standard windows modem. You want something that graphs the bandwidth through the modem at any specific time - and the ability to graph it over time. Check someplace like <A HREF="http://downloads.cnet.com" target="_new">http://downloads.cnet.com</A>, or your favorite download site.<br><br>Run this monitoring software and Ping Plotter simultaneously. Watch the bandwidth useage when your latency is bad - and see if you can correspond bandwidth useage with these periods of bad latency.<br><br>There are *so* many software packages now that use the internet as a tool. Does your stock quote program have ads on it that could possibly be being downloaded without you knowing about it? Is your e-mail program running pulling mail? Is there a application running on your PC you're unaware of? Is there any possibility of you having a virus that may be communicating with the internet?<br><br>Once you have bandwidth useage vs latency information, then you can use this information to change the way your apps are using your bandwidth, or, if this isn't the problem, you can go to your ISP with that information. If your ISP or phone company actually do have a problem providing the service you need, you'll need all the information you can possibly find to prove it to them. This includes proving that your system isn't involved in the problem - but that it *IS* their fault. Showing the available bandwidth vs. used bandwidths is a significant step toward getting this done.<br><br>Post back here when you've completed this step (with images if possible). I'll be happy to offer any more suggestions (again, remember that these are just suggestions of what I would do - not the law of how to get things done) and analysis of the information.<br><br>Good luck!<br><br>

BTW, I use a Inspiron 7000 and Windows 2000 as my main development system - and have the same modem you do. I don't see the problems you have. I've not tweaked my modem settings in any way from the standard Windows 2000 install. I get anywhere from regular 33.6ish to 50K connections to my ISP.<br><br>

I searched for a bandwidth measurement tool and found only one - DU Meter http://www.hageltech.com/dumeter which leaves a good bit to be desired. After using it for a while in combination with PingPlotter, I would note that it would be a major improvement to have the bandwidth usage info/graph integrated with PP so that response times could be more readily viewed relative to bandwidth usage.<br><br>Generally, with some exceptions, I am observing spikes in bandwidth usage accompanying spikes in poor response time. I'm not sure of the technical reasons why response time would slow under load - it's kind of like the water flowing more slowly as one opens the water tap.<br><br>