Slashdot Mirror
The FCC Says ISPs Aren't Hitting Advertised Speeds
MojoKid writes "The Federal Communications Commission has released the results of a year-long scientific study it conducted with regard to the upload and download speeds of thirteen American Internet service providers. Most of the ISPs hit 90 percent of their advertised upload speeds. Of the 13 providers tested, only four (or less than a third) averaged at or even above their advertised download speeds (Charter, Comcast, Cox, and Verizon Fiber). The tests were performed by a private firm that has run similar tests in the U.K. It measured performance at 6,800 'representative homes' nationally in March."
On my Verizon FiOS connection, I can regularly hit 25mbps on my 15/5 line for file downloads and speed tests.
I'm willing to bet that if I kept that up for extended periods it would drop down a lot, but it's fine for quickly downloading a Steam game once a month.
I know DSL, being an ATM-based technology and often subjected to PPPoE overhead, will score lower than rated. I have a 5 megabit connection but that's the sync rate. You can realistically expect to lose 9-10% just from the above overheads. That rather fits with the graphs I'm seeing.
I've seen some ISPs compensate by setting the sync rate above the advertised rate but most don't.
Verizon is evil in implementation of their contracts. They have good technology with outstanding network reliability. However, their policies and their hip-hop rapping store staff can suck it!
Comcast OTOH has good technology and a nice network too. They're pricey, but well worth the service IMHO. They customer service in India can suck my left nut, and those sub-contracted repair service men need to stay the hell away from my beer and peanut. But most importantly, they need to fix the problem the FIRST TIME AROUND. You here that Comcast, get your damn men to fix the problem. Your repeat truck-rolls are costing me time from work (to meet them) and you as well.
Life is not for the lazy.
They still are. They just hide their packet shaping from burst speed tests pretty well.
Apparently the ISP that supplies /. is one of the slow ones.
Time to offend someone
Geeze... going back in time... a 1.5 Mbit T1 connection, while actually a continuous 1.5 Mbit connection, never quite delivered that much speed when it was hooked to "the internet" and expected to move TCP/IP traffic. Same for 10 Mbit Ethernet (and that was never a true bidirectional 10 Mbits to begin with).
Protocol overhead always nibbles away at the edges.
Although I'm only one datapoint, my Optimum Boost (Cablevision) service north of NYC almost always hits the 50d/8u Mbps that I'm paying for ($15 over base service for the higher speeds). When I've had issues, they've always been catastrophic ones (no signal due to bad connector on the utility pole, etc.) rather than just slowdowns.
Aye, but there's also something this doesn't cover - region. Many of these companies perform differently if you use their service in different cities.
Where I live, you can regularly expect AT&T DSL to give you about +10% of advertised speed, and we have Wide Open West, which is good (but not listed there). The Insight and Time Warner list there probably overestimate the quality of service here where I live. Actually, since Insight (at least here) uses the TimeWarner infrastructure, it doesn't surprise me at all that they show so similar.
However, if we conversely go to a friend in another city, her Time Warner acually does meet advertised peak speed, and does so fairly regularly, but equally regularly goes down to DSL speeds.
The charts that are listed on the TFA are useless except on a per-city basis.
Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
You should send them $5 and say that you are now paying your internet bill "up to" the stated amount...
Seven puppies were harmed during the making of this post.
I have an "Up To 12 Mb" connection through my local cable company. I get exactly that. Somewhere between nothing and 12 Mb. I certainly never get more than that. But given that they advertise it as a "you might get as much as X bandwidth", I don't see how you can say they aren't giving me what they promised.
Most of the providers who hit their advertised speeds implement a burst-based traffic shaping.
For example, Comcast does full or over-full speed for first 10mb down and 5mb up.
It's nice that speedtest sites like speedtest.net show a graph of how speed changes, but their test sizes are still far to small and should exclude any detected burst speeds.
The only good way to test this is to actually transfer files and exclude the bursts.
Another thing that SHOULD be tested is the speed difference with single threaded transfers and segmented/multi-threaded transfers for both same continent and cross-seas.
Internet speed is relative and that is part of the problem.
I _can_ hit wirespeed from my ISP (AT&T DSL), but only during off-peak hours. During rushhour (late afternoon, evening, esp Sunday), I'm lucky to get ~1/4 wirespeed.
I'm sure this is AT&T overselling their infrastructure (Uverse) and has choked the uplink fiber from my DSLAM. YMMV -- not everybody will be choked. But I doubt the FCC measured this congestion.
I live in a major metropolitan area (MSP) and I have business class from Charter. I am usually at 125 to 150% of my downstream and 100% of my upstream.
Sorry it sucks for you. I've been there with many different carriers over the years (Verizon DSL, RR, ATTBI, Frontier DSL to name a few) and Charter is the most solid, fastest, and definitely has the best customer response (I only contact them via Twitter) that I have ever had before.
Granted I have business class and that may make a difference so YMMV.
When I had issues with my Comcast cable internet connection, it was taking me about 5 minutes to load Google's home page, 10 minutes for the nytimes.com home page. Slashdot took about 7 minutes. I went to three different independent speed test sites, which each confirmed I was getting less than 5% of the bandwidth Comcast advertised. I called them up and they directed me to a flash animation that looked like an analog gauge of a car speeding up onto the freeway, overshooting the advertised bandwidth, wavering a bit to make it look like it was actually measuring something and leveling off at exactly the advertised bandwidth. I reloaded it a couple times, and each time it was the exact same animation. The rep then said, "can you read me what it says on the dial? Looks like your connection is working just fine. The sites you are trying to visit must not have enough bandwidth to handle the connection." I asked if she'd ever heard of a little company named Google, and she said they must be having network trouble on their end.
The fact that they're making a "guarantee" that, in fact, guarantees nothing is in itself misleading. They're phrasing it as if it's a promise of something, when in fact it doesn't promise anything at all.
... cheat, steal, and lie. Old news. Move along.
now we need to go OSS in diesel cars
Before participation: Time Warner/Roadrunner here in Southern California gave me less than a tenth of advertised speeds. Officially 7mbits down, 1mb up, the actual service was more like 400kbits. Up to 800kb, sometimes even over a whole megabite early in the morning. (Exciting!) After the initial burst, which hit over a megabit down fairly often, there were times when it slowed all the way to single digits in KB.
Under the Sam Knows program, the FCC lets the ISPs know which subscribers are part of the test. (Bit of a problem right there, I'd say.) A few days before we had the government router hooked up, no doubt when Time Warner got word of our new "status," our speeds suddenly shot up into the advertised range. I nearly swooned the first time I saw a download go by at over a megabyte. And, interestingly enough, they've stayed there. It wasn't just some random thing. We don't usually get 7mb, but 5-6mb is the norm now.
So the info that ISPs aren't delivering stated speeds even in the FCC study is interesting, given that they seem to be jimmying the results for all they're worth.
(Speed tests before the FCC program would show us getting multi-megabits that we never saw in real life. Two things there: burst-shaping, no doubt, and I've heard that ISPs have ways of recognizing speed test traffic and giving it bandwidth.)
You don't understand. Big business has an exemption from having to comply with the law. Their CEOs have arranged for this through the Republican party.
When you stop believing that the Democrats care for you too, then you'll be able to start addressing issues, rather than tribalism.
The Democratic administration just promulgated proposed rules that would have a small farmer lose his farm if he's ever caught taking a pee on the side of the road.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
I recently watched a PBS special about broadband, which indicated the UK's system is setup so that most households have a choice of multiple broadband providers, where high speed starts in at least 2-digit megabits per second, and the monthly cost is almost trivially low (I forgot the actual costs that were mentioned).
Although the lines are owned primarily by an oligopoly of companies (AT&T, Verizon, and British Telecom were the three mentioned), they are required by law to lease the lines to competitors. Not only that, but Verizon, AT&T, and BT all wholeheartedly endorse the the concept of being required to lease their lines to competitors. Spokespeople for those companies all said that the required competition kept them working to improve their respective services.
The special also said that the companies are investing in massive new outlays of fiber optics across the country so that even very remote and sparsely populated outlying areas get fast Internet.
Now shift to the U.S., where Verizon and AT&T are fighting tooth and nail against regulations that would provide the same level of service and network expansion going on in the UK, and where 3mb DSL is considered high speed (by AT&T).
It just drove home how royally screwed we are in the U.S.