Slashdot Mirror
Thumb On the Scale? Study Finds 5 of 7 Broadband Meters Inaccurate
stox writes "For the 64 percent of Americans whose internet service provider imposes a broadband cap, and for those lucky enough to have a meter, I have some bad news. The president of the firm who audits many of the country's broadband meters says that he can't certify the measurements produced by five out of seven of his clients' meters because they don't count your bits correctly
Its all bout the money
DD-WRT has a meter I find it to be very accurate. I guess it could be used as evidence if things do not match.
If they actually use the faulty data (they paid for this for a reason, yes?) as part of their business strategy, then EVERYONE is poorly served: the company, its shareholders, and its customers.
Sounds good to me- I prefer porn with more 0's and fewer 1's.
When they were enforcing the 250 gig cap, they were within 1% of my dd-wrt tally. Now that they're not enforcing the cap, their reading is waaaaaay under my actual usage. I wonder if they're no longer counting traffic that stays in the Comcast network.
Perhaps we need a weights and measures type certification for ISPs?
In the US it's per County, so that will be interesting!
no other explanation is necessary. For the old folks here who used to have a landline phone service in the old days, do you remember all those mysterious little "charges" they tacked on your bill? Like $1.05 "User Service fee" and $0.87 "DCF Maintenance fee" or some crap like that? Well even the federal gov't realized they were just plain thieves and sued them, which they settled for a few dozen million dollars. And went right back to doing it again.
Also there was the dial-up modem scam the telcos used to pull... Dvorak's summary
So this means that they can't legally do 'metered billing,' as the meter is known and proven to be inaccurate, right?
right?
anybody?
An enigma, wrapped in a riddle, shrouded in bacon and cheese
DATA1=`grep eth0 /proc/net/dev | sed -e 's/ /:/g' -e 's/:\+/:/g' | cut -d: -f 3,11` /path/rrd/eth.rrd N:${DATA1}
rrdtool update
?
#
#\ @ ? Colonize Mars
#
Building an incorrect bandwidth meter is easy. Incorrect meters will calculate your bandwidth like ( 'MTU size' * 'number of packets' = usage), which will over estimate usage by a large margin (30% off is common), since a large number of packets are much smaller then MTU, DNS replies for example. It is 'somewhat' more accurate to take ('average packet size' * 'number of packets') per user, since different usage will come up with a different avg_pkt size. Counting each packets size and keeping track of it is the most accurate, but also the most resource intensive therefore the least likely to occur in bulk by the ISP.
Another place that can cause a significant skew in total bits is where bandwidth is monitored. Most ISPs count traffic before the restriction of your slow connection, therefore packet loss and re-transmits get counted against you (if the ISP uses no, or a bad queuing discipline this can end up being a significant amount of bandwidth). Monitoring how much was downloaded is best done on the CPE, such as the cable modem or dsl modem, but that would lead to firmware hacks and such to lie to the provider.
when I've called my ISP to complain about low speed, they usually start out by telling me to go to a specific site to check my speed. (they do the same thing when they send out a tech)
Thanks, no. I'll go to a different site. Anywhere besides the one you just suggested to me. Using what they recommend is like the used car dealer recommending you get a second opinion from his brother Jim.
I work for the Department of Redundancy Department.
If 5 out of 7 Electricity, Water or Gas meters were inaccurate, you can bet people would be screaming and government would be cracking down.
I'll see your senator, and I'll raise you two judges.
It's long been industry practice to count bandwidth as # of Packets * MTU.
It's always been bullshit, and it always will be, since the government nicely handed over practical monopolies to them
... and how much is ads?
Moving to a new client's site gave me a taste of using a browser without noscript and flashblock. I discovered a number of sites are displaying multiple ads that consist of flash movies.
To view a few paragraphs of text (a couple kilobytes or so) I USED to be downloading perhaps a quarter megabyte of graphic imagery. Now I'm downloading perhaps a minute of video for each of several self-starting video ads.
Not to mention popovers-on-mouseover - including some that that darken the whole page rather than just obscuring part of it - and if I want to kill them without "pushing a 'close' button" supplied by the popover ("Push me! Push me! I'll just close the window and not download malware! I promise!) I have to reload the page all over again. Listen to that meter whir!
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
I don't know how many on slashdot know networking, but there are different ways of measuring bandwidth. Are we measuring layer 2, or layer 3? Further, layer 2 can often have multiple encapsulations before even taking layer 3 into consideration. Take for example DSL which frequently uses PPPoE, which means we have both PPP and Ethernet frames in addition to the IP data and everything encapsulated therein. And if you include DSL interleaving, then do we also include the packets that had a bad checksum and were therefore discarded? (in many cases there are a lot of these) That *is* data usage by all definitions. Do we also include ingress packets that were dropped due to bad checksums? Again, that is data usage.
In my opinion, the problem is that there aren't any standards defined for measuring bandwidth. Also in my opinion, that definition should be layer 3 traffic only and nothing else.
Careful with names containing L slashdot.org/~AiphaWolf_HK slashdot.org/~AlphaWoif_HK slashdot.org/~AiphaWoif_HK
I guess it really doesn't take any facts for the idiots to start clamoring about how all business' are evil.
Don't complain about syntax, grammar, or spelling. There is no.hell like input on android.
If 5 out of 7 Electricity, Water or Gas meters were inaccurate, you can bet people would be screaming and government would be cracking down.
Also, imagine if the cost (to the provider) of providing Water at near-capacity was roughly the same as the cost of not providing any water, because the bulk of the expense is in maintaining the pipes... Imagine the outrage against the greedy provider who strictly limits the amount of water anyway.
That's called "ebony."
I'm sure these meters make mistakes both ways right? Occasionally under counting.
From the article:
"They are wrong by missing numbers by one way or another - sometimes it's over reporting, but more frequently the error is under reporting," he said. Under reporting should be a relief to those facing overage charges or service termination for going over their meters, but if the meters aren't counting the data properly, it is still a problem.
Does NOBODY RTFA anymore? Sheesh....
imagine if the cost (to the provider) of providing Water at near-capacity was roughly the same as the cost of not providing any water, because the bulk of the expense is in maintaining the pipes
Sounds like reality to me, what are we imagining again?
Imagine the outrage against the greedy provider who strictly limits the amount of water anyway
Sounds like reality to me, what are we imagining agian? (Seriously, take a look at any place in the US that rations water during droughts, you'll find people watering at 3AM to not get caught cheating, and other people calling the cops at 3AM out of spite). The only difference is that the ISP can turn off the packets coming to you from their end, while the water company can't turn off the molecules of water coming out to your house without sending a truck.
Bandwidth not used is lost forever. Water, not so much.
WTF kind of pages are you viewing that would even dent your limit? Or what kind of account, perhaps? I'm on the most basic shaw.ca service with an internet cap of 125g/month.
Heck-yes, overloaded pages are annoying, but unless you've got new information, we're talking about downloading and streaming big files. Your pet-peeve isn't relevant.
If they send you 1 octet, and have to re-transmit it 5 times in the link layer, how many billable octets of data have you been sent?
Thats right: layer2 can multiply your data, and you can be billed by some ISPs for octets injected, which requred re-generation inside the system.
how long is a byte? its not 8 bits. sometimes its 10. If your link layer adds bits, these count at bit-counts towards a higher octet count.
Do you think anyone would know? How many people have real clue how to measure how many kw of electricty they use, or how many gallons of water flow through their house each month?
I have no idea how many gallons of water I actually use, my bill could be off by a factor of 10 and I wouldn't be able to tell.
And truthfully I wouldn't really care, what I care about is if my bill is in line with others that appear to use the same as me.
They can call the units anything they want to call them as long as I can get a reasonable estimate of how much more my bill will go up when I use twice as much..
The bytes counted by my router are not the same as the bytes counted by Comcast, that's for sure. I wonder if this means I can SYN flood someone for free?
Water, Gas and Power meters are required to be required to be accurate by federal law. These laws included the establishment of agencies like Underwrites Labratory to certify power meters.
You know all those laws and regulations your parents enacted to prevent people from cheating and keep the system as fair as possible, that certain people today think should go away because all government regulation is bad. Where the reality is they are trying to propagandize people who aren't smart enough to realize what they are suggesting is going to result in the wholesale ripping off of the poor and middle class just like it did before these laws were enacted.
Who the hell mods a 0-score post as "overrated"? And get your sarcasm meter checked, it's time for a humour top-up.
Someone who thinks it should be at -1 even though it's not troll, offtopic, redundant, or flamebait?
For example, wrong information, or a superbly silly statement/question that still manages to apply to the topic. Not that this applies here, I'm just answering your question.
For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
Go buy yourself any telco bit pipe. You're always charged based on the size of that pipe regardless of overhead of any protocols used over it. So if your ISP sells you a 10 mbit connection, but requires your router to run PPPoE then that means for every 1492-byte packet of IP data, you're going to have 8 bytes of PPP overhead and 18 bytes ethernet overhead for that packet that count against your bits used, ignoring the TCP/UDP packet header overhead or any application level protocol overhead. Yep, that's just the way it works. And don't forget that your router and theirs might also be sending out ARP or other periodic ethernet proto traffic (depending on how things work) that will ALSO occupy bit space on the pipe.
I'm unaware of any ISPs where I live that only bill/serve based on the IP throughput of a pipe. Every single one I know of bills or sets service levels just like any telco would, based on the raw bit pipe size.
Actually, this isn't exactly true. If everyone jumps on and watches video in the evenings, then the local shared pipe will get saturated and everyone's experience will suffer. Running more lines could solve the problem. So, in essence, bandwidth could "be full". But that's not what's happening, except in rare instances.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
In other words, the amount of data layer 4 can send/transmit in a given unit of time.
Layer 3 is nearly as close, but it may include more data that's not inherently "usable" but necessary (TCP overhead, ICMP packets, etc).
Layer 2, with its many levels of encapsulation on the ISP side is less valuable because it involves so much overhead.
How about some honesty in bandwidth numbers?
Comcast is what I'm thinking of specifically -- they provision your *modem* to talk to the local head end at ridiculous bandwidth numbers but in my experience, once you go over about 25 Mbps in most areas you never see it, even if you are provisioned at 50 or even 100 Mbps.
I've been caught in the middle with customers who have equipment provisioned at the 100 Mbps level before who see nowhere near this, even on Comcast's own cheesy bandwidth meter. The customers insist something is wrong with their equipment (since this is what Comcast suggests), and trying to explain the nature of Comcast's network and where bottlenecks occur isn't easy (especially when you're making assumptions of your own).
Personally, I think ISPs should be required when selling a given bandwidth tier to actually label it based on the ability to burst for at least 5 minutes at the maximum throughput level @ layer 3 during ANY time of day to an off network peer.
It's completely dishonest to sell 100 Mbps of throughput when you only have the local upstream backhaul provisioned at that speed and it's shared by many people.
It is quite likely that they are exporting Netflow from their routers. Netflow will take "snapshots" of the data from hosts, but isn't 1:1. Typically, they're only export Netflow at a 1:6000 rate, which means that one out of every 6k packets are sampled. From here, they extrapolate the estimated bandwidth consumed. It's generally not 100% accurate, but it'll get you into the general ballpark of the real usage.
tl;dr:
They're likely extrapolating your bandwidth based off of sampled metrics, and it is not 100% accuratehttp://tech.slashdot.org/story/13/02/07/2137247/thumb-on-the-scale-study-finds-5-of-7-broadband-meters-inaccurate#