This is for anyone that is still a part of the SamKnows/FCC study on ISP broadband performance. I left the study at the end of last week because the NETGEAR router they give us back in 2011 was exacerbating our problems running up against Centurylink's 250 GB bandwdith cap.
I've spent the last week tracking down some unaccounted for bandwidth usage that I started noticing between the hours of midnight and 8 AM, a time of the night when I'm up and active but nobody else in our household is. I would see as much as 200 MB of use (according to the NETGEAR router) between midnight and 2 AM, most of which I couldn't account for based on my own usage. This would continue into the morning and *seem* to taper off after 8 AM, but I couldn't be sure if to slowed, stopped, or even continued unabated.
Finding the source of this phantom usage wasn't an easy task. We have six people living here which means six computers and eight non-PC Internet-connected devices: Xbox 360, iPad Mini, Nintendo 3DS, iPod Touch, Kindle Fire, an ASUS tablet of some sort, and two Dish Newtork satellite receivers.
Tracking usage on the PCs was very easy, a program called BWMeter keeps a very accurate record of usage on each computer and allows the stats to be read remotely. On a sample day (July 5th) when my nephews weren't home (they are home for the summer from school which is making staying under 250 GB a ridiculous task), I was able ascertain the following:
1. The NETGEAR router reported 1595 MB of downstream usage in that 24-hour period.
2. All computers combined used 551.6 MB of downstream.
3. All non-PC wireless devices were off that day, leaving only the satellite receivers.
At this point I used a setting on the NETGEAR/SamKnows router to disable the SamKnows tests for 24 hours covering July 4th and July 5th each, and dismissed the router as the cause when the phantom usage appeared on the 5th -- the unexplained 1 GB or 65% of our total that day. That only left the satellite receivers to test.
I knew that both receivers could stream TV and movie content on-demand (and that nobody ever did that) from our broadband connection, but little beyond that. I thought I heard somewhere that they will use your broadband connection to report PPV purchases and download the program guide. Disabling them was easy since these models use one of those adapters that turns your home's electrical wiring into a slow LAN. I unplugged that adapter from the router and waited a few hours and still saw phantom usage.
The most likely explanation -- with every device in the house ruled out -- was that the NETGEAR router was still doing tests even though I told it not to. I tested that theory by disconnecting the GigE switch that all our PCs use from the router, and by shutting off every single wireless device for a set period of time over night while everyone but myself was asleep. At this point the only thing connected to the DSL modem was that NETGEAR/SK router, and nothing was connected to the router.
I connected my laptop to the router via wireless briefly to check usage at 3:08 AM on July 5th, and it read 220 MB down. I confirmed via the "Attached Devices" page that no other wireless devices were active, and then I shut the laptop off and waited almost an hour. I checked our usage again at 4:00 AM. Usage had ticked up to 233.79 MB, an increase of 13.79 MB which could only have come from the router itself. Less than what I'd expected, but I also noticed from my testing that the phantom usage tended to come at the top of the hour. I shut my laptop off and waited another ten minutes (4:00 to 4:10a), and then checked again. Usage had increased from 233.79 MB to 265.99 MB, an increase of 32.2 MB.
45.99 MB of unaccounted usage over the course of 62 minutes might not sound like much, but what you've got to understand is that this is happening every single hour of every day, all month long. The faster your connection is the more data these tests use, and it's not always 45.99 MB.
I recorded the bandwidth usage according to the router and also according to every PC on our network (with all other devices shut off) for 24 hours. The results are in the attached image.
The first column is the date and time that the numbers were finalized, the "Total" is downstream usage as reported by the NETGEAR/SK router (cumulative), "Diff" is how much downstream usage there was for each hour, "Computers" was how much data was used by the devices that I could directly measure, and "Unaccounted for" is self-explanatory.
For 2 AM on July 5th, you can see the router reported a total downstream use of 153.33 MB, but only 28.47 MB used between 1 AM and 2 AM, of which the computers used only 3.8 MB, leaving 24.67 unaccounted for.
There are several obvious patterns to the unaccounted usage which is more evidence that it is indeed the result of SamKnows tests run by the router. The unaccounted for usage marked in blue, between 13 and 15 MB, occurs every fourth hour on the dot. It is always followed by use between 60 and 67 MB, and sometimes it's that much twice in a row. In one instance, 1 PM through 3 PM, the unaccounted for usage totaled 188.46 MB.
Again, if you think that's not much, look at the unaccounted for total for the day: 1.01 GB. If that one day is representative of what a days worth of tests normally looks like, then the NETGEAR/SamKnows router -- for us -- is using slightly more than 31 GB per month, or 12.4% of our cap.
Our line is 6mbit, so people with a faster connection (DSL or not) will see even higher usage than this. We've been living with a de facto 220 GB cap since last year (when we got bumped to 6mbit from 3mbit), and because we're up against 250 every month, this situation is no longer tolerable for us.
I explained this situation to SamKnows and asked them to release me from the study and they quickly and politely obliged. I offered to return the router to them on my dime, but they said they only require returns from the UK, and that I should dispose of the router in an environmentally friendly manner. I tried to flash the router with the most recent firmware from NETGEAR, hoping I could wipe out the tests, but it wouldn't take. I simply bought another identical router from Newegg because I really like this model and it was cheap ($45 or so).
I expect to save about 31 GB per month by leaving the SamKnows study. If you're a part of this study and have a bandwidth cap, I suggest that you do the same.
↧