Many thanks to Doug for sharing his trials and tribulations from the networked netherlands.
It took some patience and perseverance, but I did get some results.
My English-speaking tech was kind of a dead end. His job, and his entire focus, is working with cables, connections, modems, and routers.
"If your only tool is a hammer, everything looks like a nail." His focus was on what he was used to fixing (noise in the line, bad connection on the pole outside the house, reboot the modem and router, etc.) and I could not get through to him that the problem was not local.
Why did I think this? Because even when pinging Google showed absolutely no connectivity, I could always ping the gateway IP address and get practically instantaneous (<1ms) returns, and the cable TV never faltered.
After presenting him with a several printouts demonstrating this, he turned me over to the engineer at the office. He had about the same amount of English as I had Spanish, so we were able to more or less communicate, and he found the printouts informative and intriguing, although for the first two weeks he was stuck in the same rut as the installer, looking for local fixes. He, too, gave me his cell phone number!
I don't know if it was coincidence, or if the engineer finally put pressure on upstream (where I believe the problem originated) but at 3:30 (more or less) on October 22, following a nine and a half hour outage, the internet came back on, but with a difference. The ping times which had been running (on the occasions when there were any pings at all!) an average of 50-60 ms, were now running 70-90 ms, about a 50% increase, but still perfectly acceptable. The big difference, however, was an increase in reliability, from about 50% when I started logging to more than 97% since October 22. It would be more than 99%, except there was a [probably] planned outage of 2.5 hours on October 25. Other than that outage, all of my glitches have been single, occasionally two in a row, lost packets scattered randomly maybe 10 times per day. I'm told this is normal.
My thinking is that there was a problem with some piece of equipment upstream from the ISP, evidenced by gradually increasing ping times getting up into the 300-400 range before total loss of connectivity; then sometime later, frequently at the top of the hour, the service would return rock solid (for a while) with pings in the 50's, then after several hours (sometimes just a couple, sometimes 15 or 20) the cycle would repeat.
I think someone, somewhere upstream, was being reminded by their watch beeping on the hour, and going to "check the line" and resetting something. Finally they either replaced a failing piece of equipment, or possibly changed the routing so the ping to Google.com takes some extra hops, which accounts for the 50% increase in ping times, and the huge increase in stability. Of course, this is just surmise on my part, based on invincible ignorance of how the internet actually works, but it seems plausible to me.
I am now pressing the ISP to give me six months of free internet service as compensation for several months lack of service (the problem was there for at least one month before I began logging, possibly more) and for the thorough documentation I provided. The engineer has agreed that this is reasonable, and said he would work on it. I'm not holding my breath.
Since I doubt that anyone here is interested enough in my tribulations to want to look through 51,618 data points (that's what I've collected so far) I am attaching a screen shot, showing just the statistics for the first week and the last week (more or less) of the data collection. And if there are any Excel freaks among you, I am attaching the Excel template into which I paste the CSV data from the PingLogger batch file. There are a couple of neat things in it, particularly cells D2 and F2, and the conditional formatting in column C. [Owww, my arm hurts from patting myself so much on the back
]
tanstaafl.
Previous Topic
Index
