Netflix Shows Which ISPs Actually Perform Well... And Which Don't

from the ah,-data dept

Fri, Jan 28th 2011 3:02pm — Mike Masnick

It's always fun to discover some data that reveals some useful info -- and the folks over at Netflix have apparently realized that they're sitting on a treasure trove of data concerning the performance levels of various ISPs, based on all that content Netflix is streaming. Rather than just keep it all internally, Netflix is starting to publish the data, showing how well certain ISPs perform, and highlighting how you generally get better performance out of cable than DSL.

This is smart on Netflix's part, as it actually does a nice job encouraging ISPs to improve their quality, knowing that they can tout these kinds of "independent" rankings to show their overall quality.

Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.

Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.

While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.

–The Techdirt Team

Filed Under: bandwidth, cable, dsl, internet
Companies: netflix

41 Comments

If you liked this post, you may also be interested in...

Reader Comments

Subscribe: RSS

View by: Time | Thread

Anonymous Coward, 28 Jan 2011 @ 3:07pm

Unless they have exactly the same peering to all ISPs, the nubmers are at best misleading. The issues to some ISPs may be that Netflix doesn't peer properly.
[ link to this | view in chronology ]
Vinny, 28 Jan 2011 @ 3:16pm

No Way
No one that looks at this chart should take it seriously. Fios smokes every single cable provider in speed and Verizon is at the middle of the pack?

Oh wait... Not Fios. Verizon... Right...

So they averaged out Fios and Verizon DSL. Got it.

As someone whose dealt with Netflix on Cablevision and Time Warner there's now way in the coldest reaches of hell that Fios isn't at least twice as fast as those two. They should've separated Verizon into two parts and the people posting this should've known better.
[ link to this | view in chronology ]
- Anonymous Coward, 28 Jan 2011 @ 3:28pm
  
  Re: No Way
  How do they separate the two? DNS?
  [ link to this | view in chronology ]
- Anonymous Coward, 28 Jan 2011 @ 5:07pm
  
  Re: No Way
  No one that looks at this chart should take it seriously. Fios smokes every single cable provider in speed and Verizon is at the middle of the pack?
  
  FIOS generally has the *capability* to do so, but that doesn't mean that the provider will provision it to actually do so.
  
  Oh wait... Not Fios. Verizon... Right...
  So they averaged out Fios and Verizon DSL. Got it.
  
  And you know what else? It varies by city and neighborhood and tier and sometimes even by individual customer, too! You know, some customers have faster computers than others! So what do you want, for them to publish numbers for each and every customer? It that the only way you would "take it seriously?" No, the chart is exactly the way they claimed it to be, i.e. per ISP. Not per tier, not per city, not per neighborhood, not per customer, but per ISP.
  
  Sorry if that isn't good for some of your stocks.
  [ link to this | view in chronology ]
  - Greevar (profile), 28 Jan 2011 @ 6:57pm
    
    Re: Re: No Way
    Faster computers? Did you actually claim that as a contributor to network performance? If you have a computer that would degrade your network performance, it must still be running on a Pentium II.
    [ link to this | view in chronology ]
    - Anonymous Coward, 28 Jan 2011 @ 7:15pm
      
      Re: Re: Re: No Way
      Faster computers? Did you actually claim that as a contributor to network performance? If you have a computer that would degrade your network performance, it must still be running on a Pentium II.
      
      I've got a 2 GHZ Pentium 4 on which I guarantee you that if I'm running some CPU intensive applications in the background that it will slow down streaming video. Think before you speak.
      [ link to this | view in chronology ]
      - Anonymous Coward, 29 Jan 2011 @ 9:22am
        
        Re: Re: Re: Re: No Way
        The slowdown is inside your box, not the network. Take your own advice :)
        [ link to this | view in chronology ]
        
        Anonymous Coward, 29 Jan 2011 @ 7:58pm
        
        Re: Re: Re: Re: Re: No Way
        The slowdown is inside your box, not the network
        
        Umm, who said otherwise? But from what Netflix sees on their end, it is still a slowdown and affects the numbers in the same way.
        
        Take your own advice :)
        
        Me thinks you are the one who should perhaps take a little of it (in order to keep from appearing so foolish).
        [ link to this | view in chronology ]
- Patty, 29 Jan 2011 @ 2:54am
  
  Re: No Way
  I agree. I get astoundingly fast speeds with my Verizon FIOS and absolutely no problems. I signed up as soon as it was available in my area, one of the earliest to be activated. I had Adelphia, then Comcast, then Verizon DSL. I would never ever consider going back to cable - the slow downs, the down time, the incessant and never ending calls to tech support, a never ending nightmare, no exaggeration. FIOS is a dream, a wonderful, wonderful dream.
  [ link to this | view in chronology ]
- Mr. Oizo, 29 Jan 2011 @ 10:11am
  
  Re: No Way
  What do you mean with 'average them out'. It makes no sense what you say, because you get the actual numbers not the averages accross all ISP's. So, could you please explain this in detail ?
  [ link to this | view in chronology ]
  - Cynyr (profile), 30 Jan 2011 @ 9:42am
    
    Re: Re: No Way
    the verizon line includes both the speeds the FIOS people get and the people paying for the slowest tier of DSL. hence "averaged out"
    
    the Comcast line is the same way, people in one area may have a max of 50Mbps, while across the country the max speed may be 25Mbps. This chart does not tell you if in New York City Charter is the best ISP for streaming netflix.
    [ link to this | view in chronology ]
- leichter (profile), 30 Jan 2011 @ 6:42pm
  
  Re: No Way
  If you're going to split Verizon, you have to split AT&T along the same lines - classic DSL vs. U-Verse, with the slowest marketed U-Verse speed being twice the highest classic DLS speed.
  
  But if you start down that path, why not start splitting cable providers into areas where they've built out DOCSIS 3.0 and areas where they haven't?
  
  The fact is that no one (other than the carriers themselves) have data this fine-grained - and it's not even clear that they do. Oh, they know the theoretical speeds of their "last-mile" offering; but that's nowhere near the same thing as actual measured data. Fios, for all the theoretical speed of its fiber, could have crippled connections into and out of wherever all that fiber converges. Or there could be other things limiting them, perhaps even stuff they aren't aware of.
  
  The Netflix data is valuable exactly because it's measured, not a guess. Networks are complex combinations of many different components, and you're kidding yourself if you think you can know the performance of any non-trivial network without measurement.
  
  ��-- Jerry
  [ link to this | view in chronology ]
eo, 28 Jan 2011 @ 3:24pm

Smart move
As ISP's around the globe starts to charge for traffic, Netflix takes them by surprise, making the bw-limiters suffer!
[ link to this | view in chronology ]
trilobug, 28 Jan 2011 @ 3:53pm

I really hope this wakes people up.
You should post the Canada Stats - their lowest is right where our highest is.
[ link to this | view in chronology ]
- freak (profile), 28 Jan 2011 @ 7:37pm
  
  Re: I really hope this wakes people up.
  Yes, but we have to deal with throttling after a hard limit, or a usage cap.
  
  Seriously, I have only have 2 ISP choices, and one throttles me back to about 2x dial-up speed after 60GB per month, and the other, although much faster, charges me $5 per GB after the first 50GB.
  
  That's for the $120/mth plans.
  
  Like down there, it's much better near the larger cities, but around the medium-sized ones, well, it's crap. But at least, unlike my parents, I'm not stuck on dial-up.
  (Also, some easy router hacking will feed them incorrect information, so their counter never goes beyond 59GB, so it never throttles, but still!)
  [ link to this | view in chronology ]
fogbugzd (profile), 28 Jan 2011 @ 4:08pm

Canadians
There are lots of problems with the data, but those problems don't mask what I think is the big reveal from the charts.

Looking at the original article one thing that strikes me is that all of the Canadian suppliers except one are significantly higher than the all of the US providers, and the lowest Canadian speed is right up there with the top tier of American suppliers.

This shows how poor competition is in the US. US providers often say that US service is slower than other countries because we have low population densities. Well, Canada is even has even lower population density than the US, even if you only consider limit the calculations to 200 miles of Canada's southern border.
[ link to this | view in chronology ]
- btrussell (profile), 29 Jan 2011 @ 5:26am
  
  Re: Canadians
  Canada isn't faster, just more consistent.
  I've downloaded 4GB files at an average speed of just under 1MB/s(1017KB). This average includes 4-5 minutes of disc space allocation time for the file. This is for a DVD .iso. LiveCDs', being much smaller files and taking less time to allocate space, give me higher download averages.
  My connection speed is 14Mb (1.75MB) down, 136KB up. The fastest connection I can buy here is 16Mb/s.
  Some peers from US upload @ 1MB/s.
  [ link to this | view in chronology ]
Anonymous Coward, 28 Jan 2011 @ 4:17pm

Sigh. The color choices suck. I can't tell if I should tell Comcast to bite me or not or if it's Cablevision.

Fair enough to tell inComPetentCast to bite me? Past history says 'YES!'

- tech sitting in my living room for 3 hours waiting on Comcast dispatch to approve my already paid for HBO? check.
- nearly a month to get a working HD box and having to return a STACK of non-functional ones on my own time? check.
- calling to find a solution to my 'decided to rename your wireless network so that you can't use it anymore' and being told my router modem is no longer supported except it's not that at all and I found the fix on the internet somewhere else within 5 minutes? check.

Yes. Bite me, Comcast. Bite me and choke.
[ link to this | view in chronology ]
- Eugene (profile), 28 Jan 2011 @ 4:23pm
  
  Re:
  Yeah, I don't think you're going to see 2 week-long outages due to Comcast's gross incompetence on this graph. Netflix can't see date if people aren't even able connect in the first place. :P
  [ link to this | view in chronology ]
  - Anonymous Coward, 28 Jan 2011 @ 5:35pm
    
    Re: Re:
    Yeah, if they factored in outages, there's no way in hell Charter would be #1
    [ link to this | view in chronology ]
James, 28 Jan 2011 @ 4:38pm

That's hardly fair to Verizon; cable is better than DSL, but FIOS is a million times better than cable. If they broke down Verizon FIOS and Verizon DSL, the DSL would be lower than the current red line and FIOS would be a mile above every other color on the chart. :)
[ link to this | view in chronology ]
Steve, 28 Jan 2011 @ 4:41pm

Netflicks Internet Provider Rankings
Clear works better for me with Netflicks than Comcast did.
[ link to this | view in chronology ]
Memyself, 28 Jan 2011 @ 4:49pm

I'm very interested in this chart, but I'm just colorblind enough to not be able to read it properly.
[ link to this | view in chronology ]
Anonymous Coward, 28 Jan 2011 @ 6:01pm

Now I have proof that my ISP sucks.
[ link to this | view in chronology ]
Mike Doherty, 28 Jan 2011 @ 7:28pm

Google's HDD study
This reminds me of Google's study on HDD failure rates. It was a great study, because Google is such a massive consumer of so many different makes and models that you can get very good statistics out of the data. But the study was flawed in one critical way -- they didn't name names because the failure rates are considered "proprietary knowledge." So kudos to Netflix for doing a good job of collecting, analyzing and presenting the data *including* identifying who's who.
[ link to this | view in chronology ]
Ryan Diederich, 28 Jan 2011 @ 7:32pm

Fios
Fios may have the potential to go much faster, but it fails for an important reason. It isnt in place everywhere.

Even if your entire town has a fiber network, as many towns currently do, that doesnt mean your provider does, or that it doesnt hit normal lines along the way.

The speed is dictated by the lowest speed encountered on the network.
[ link to this | view in chronology ]
Sully, 28 Jan 2011 @ 7:44pm

Very interesting results
This is showing average Netflix user performance by ISP. I find it very interesting for a couple reasons. On average, based on Netflix perf numbers, Comcast, the company that brought the country (or at least the tech community) into an uproar by limiting bandwidth for a specific type of download, has the second highest average on the chart. THe other funny thing is that despite this, people still want to for some reason complain about comcast.

In the context of overall service, feel free to complain about Comcast if you feel your service from Comcast has been bad. In the context of this article and this specific data point, for US ISP's, you have to say that Comcast is on average the second best performing ISP.
[ link to this | view in chronology ]
- Anonymous Coward, 29 Jan 2011 @ 4:01am
  
  Re: Very interesting results
  Comcast, the company that brought the country (or at least the tech community) into an uproar by limiting bandwidth for a specific type of download, has the second highest average on the chart.
  
  Yes, but remember that the chart is for Netflix streaming video only. That's not all there is to the internet.
  
  THe other funny thing is that despite this, people still want to for some reason complain about comcast.
  
  Maybe those are the people who use their internet connection for something besides just Netflix.
  
  In the context of overall service, feel free to complain about Comcast if you feel your service from Comcast has been bad.
  
  Oh gee, that's so kind of you to give your permission.
  
  In the context of this article and this specific data point, for US ISP's, you have to say that Comcast is on average the second best performing ISP.
  
  Unless you use it for some other types of downloads, that is.
  
  Do you work for Comcast, have stock in them or what?
  [ link to this | view in chronology ]
Dohn Joe, 28 Jan 2011 @ 9:12pm

Answer to Net Neutrality
This might just be the answer to the net neutrality debate! If "content providers" can supply data showing ISP performance then any "throttling" to that provider's content will stick out like a sore thumb and, with any luck, the internet users who are fans of that site will drop a bad ISP like a spicy Mexican-food deuce!
[ link to this | view in chronology ]
- Anonymous Coward, 29 Jan 2011 @ 9:00am
  
  Re: Answer to Net Neutrality
  You make the presumption that there are options to choose from. In too many parts of the US, regardless of size, population, etc., there are perhaps all of 2 choices of ISP, both charging basically the same amount for the same stuff, raising prices in tandem, using the same cost appropriations, the same insistence on bundling, the same introductory offer pricing shell games...customer service may be the only differentiating factor along with speeds but the average schmoe will only have customer service to go by since neither ISP will or has yet to make speed determination tools easily available (or understandable).
  
  Internet service should be declared a utility like phone, water, power, if only so we can actually see what we're supposedly getting for our money as far as true usage and speed.
  [ link to this | view in chronology ]
Anonymous Howard, 28 Jan 2011 @ 9:43pm

"clearwire" should just hang it up.
[ link to this | view in chronology ]
harbingerofdoom (profile), 29 Jan 2011 @ 12:18am

while i know that not everyone in comcast land is happy 100% of the time (or hell, even 10% of the time) i do have to say that in the past year, i have only had to deal with one major issue with intermittent outages that was a local network issue that they had to roll a field engineer on. in my experience, its not their network that drives most people to scream about comcast, its their shoddy customer service and support (id rather smash my toe with a ford falcon than talk to comcast support, fortunately i rarely have to talk to them).

this graph does not take anything beyond network performance broken down by ISP... how well they perform when you have to call them is a completely unrelated issue.
[ link to this | view in chronology ]
- Anonymous Coward, 29 Jan 2011 @ 2:17am
  
  Re:
  Seriously, in Japan in ten years I never called customer service not once. Maybe I'm lucky, but I can tell you that I changed service providers 3 times and I could choose from a list of more then 50 ISP's available.
  [ link to this | view in chronology ]
Anonymous Coward, 29 Jan 2011 @ 1:22am

This is one of those times that I thank God I don't live in the U.S.A.

Think about it, when people start passing medical data through the internet that would mean Gigabytes of information traveling to and from some point, 60 gigs will be something to laugh about it, speed like 20Mbits/s are laughable and not near enough to supply people with what they will need and the throttling and separate tiers will make your medical bill even more expensive because it goes on top of all other problems that the U.S. face in that sector, your business that needs that infra-structure will be taxed also, if you telecommute you will be paying more also, heck those people are eager to get that money they just aren't thinking about others.
[ link to this | view in chronology ]
Anonymous Coward, 29 Jan 2011 @ 1:28am

Also this is a reason why some European countries and most Asian countries will beat the U.S. again.

They are building their infra-structure and keeping it open, so when that workforce starts to get online and services like medical assistance, education, accounting, maybe even legal advice can be done remotely from India,start to appear you can bet that Americans will find themselves in a very dark place to compete with all the other players.

What will happen to all those white collar jobs when others are being competitive online and they cannot learn how to do it because somebody had the bright idea of limiting usage because of greed not technical problems?
[ link to this | view in chronology ]
- Rich, 29 Jan 2011 @ 5:09am
  
  Re:
  Well, it's good to have a dream. So, you just keep on holding your breath and wishing with all you might. Maybe Santa will come this year, too.
  [ link to this | view in chronology ]
Jon Renaut (profile), 29 Jan 2011 @ 6:43am

Who cares? There's no competition
At my house, I can get Comcast cable or Verizon DSL. None of the others on the chart are an option. And I live in Washington DC, hardly an underserved backwater.

I like to see this data come out, but I think it's pretty unlikely that it leads to anything but a couple of blog posts.
[ link to this | view in chronology ]
Hephaestus (profile), 29 Jan 2011 @ 10:18am

weird thing ....
The telco's refuse to tell the governmment which addresses they provide service to. Perhaps netflix could also publish street address ranges and which providers service them.

They could refer to it as monopoly mapping.
[ link to this | view in chronology ]
Cynyr (profile), 30 Jan 2011 @ 9:51am

graph of connection technology performances
[ link to this | view in chronology ]
Cynyr (profile), 30 Jan 2011 @ 9:57am

graph of connection technology performances
This graph basicly looks like a chart of the differances in speed between, Cable, DSL and Satellite. Sure you have some slightly faster cable providers, but in general they all are about the same speed.

This also isn't accounting for anyone QoS'ing their own connection, so that watching 3 netflix streams doesn't cripple the internet.
[ link to this | view in chronology ]
Carl Cravens (profile), 31 Jan 2011 @ 7:59am

But what rate is the customer buying?
The big factor this chart is missing is the average rate the customer is paying for. You don't typically buy cable Internet when you want a cheap connection, you buy DSL. ATT's average shown here is right around 1.5 mbps... the max rate of their cheapest plan. If a majority of ATT subscribers using Netflix have 1.5 mbps connections, then ATT's average on this chart is going to be low... even if every customer is getting every bit of speed they're paying for.

This chart is meaningless for determining how fast a given provider is at a particular speed/pricing tier. "Cable is faster"? How about "Cable subscribers pay for higher bandwith"?

I pay ATT for a 3 mbps connection, and I get connection rates with Netflix that reflect that. (After ATT got their hand slapped for throttling, that is.)
[ link to this | view in chronology ]