Tuesday, December 1, 2009

U-Verse reliability update and some cable internet geekery

A couple quick updates...

U-Verse reliability

You might recall that after almost three flawless months, our U-Verse turned to crap in mid-November, going down for four hours on the 17th and having major line noise issues the following night.

We called AT&T for service, and they were unable to find any issue with the line the next day when they were out at our house, although the record from the previous night indicate some line noise. They did do some "wave the dead chicken" troubleshooting by switching our port on the VRAD and providing us with some noise filters for the ethernet cables.

Did those make a difference? It's been almost two weeks since then with no additional downtime. Of course, those two nights could have had nothing to do with us, but rather AT&T head end problems. We know the 4-hour outage was a higher-up issue as U-Verse service thorough the KC region was affected.

Oh well. After three full months of U-Verse, our uptime is at 99.74% (it was 99.99% before the glitches). Sunflower Broadband, for reference was at 99.14% for the first 9 months of 2009.

Cable Bandwidth Meter Geekery

Those of you on Sunflower have bandwidth limits, which you can track using a meter provided by the company. Have you ever wondered how the meter works or how accurate it is? Comcast, which is experimenting with similar metering, commission an audit of its meters by a third party, which can be read right here.

The summary? The Comcast meter is accurate within about 1% of actual use. FWIW, I don't know if Sunflower's meter uses the same technology as Comcast.

As an aside, one of the major problems with meters is that there's no real-time way to know if you (or something on your network you don't know about) is sucking down the bits. Sunflower has improved their meter to the point where it shows hour-by-hour usage, which helps somewhat, but what is really needed is a real-time "tachometer" to see the actual bandwidth. This probably needs to be built-in to the router or cable modem. Ironically, U-Verse's home gateway includes a built-in meter exactly like this...I say ironically because U-Verse doesn't cap bandwidth, so the meter is informational only, although still quite useful for figuring out of someone in my household is killing our connection by hogging the pipe. If you have U-Verse you can see the meter at


  1. 99.14% over 9 months amounts to about 2 1/3 days offline. If you look at 99.74% reliability over 9 months that amounts to about 3/4 of a day offline. Are you really trying to convince me that 1.5 days over 9 months is significant?! You have got to be kidding me if that matters at all! To answer your question, yes, sunflower uses the same technology as comcast for bandwidth metering.

  2. Thanks for your comment!

    That is great that Sunflower uses the same technology; that increases my confidence in the accuracy, although Sunflower, like Comcast, should hire a third-party auditor to ensure this.

    And yes, 1.5 days over 9 months is significant. Divide that out, that would be the equivalent of 4 hours a month, or an hour a week. Would you accept an hour of downtime of the cable television signal every week (or 15 minutes a night over 4 nights), especially if it came during prime time and interrupted your shows?

  3. You make a good point that if the outage occured every night during prime time it would be annoying, but we are talking about internet outages not video outages. 15 minutes of internet outages a night fails in comparision to 15 min of video outages every night. Also, I am sure that most of this internet downtime resulted from one or two planned maintenance events done during in the wee hours of the morning to minimize customer impact. I have had Sunflower Broadband service for many years and have maybe once been upset with an outage in my area and according to Sunflower it was related to some idiot who crashed his car into a cable pedestal.