I take back everything I've ever said about BT Broadband.
Content streamer Netflix has released the first in a planned monthly listing of the fastest internet service providers operating in the US, and so far Google Fiber is the top dog. That's good news for the residents of both Kansas Cities and a lucky few in Stanford, California, but not a lot of use for anyone else. Second on …
I take back everything I've ever said about BT Broadband.
Thats NOT the average internet speed. Its the average speed for watching netflix movies
The users could be doing other browsing at the same time and the netflix service might top out at about 2.5-2.6mbs anyway.
The importance in the figures is the difference between the ISP's when using the netflix service/
I used to be somewhat envious of the Americans, back when UK Internet connectivity was via dial-up ISPs that charged phenomenal amounts, plus telephone providers that raked in the call costs.
How times change.
Probably the free tier of Google fibre is used by too many people.
"Up to 5Mbps download, 1Mbps upload speed • No data caps • Free service guaranteed for at least 7 years • Includes Network Box"
It is a bad study.
(Even on the 1G tier Netflix won't be able to max that out for every customer and won't need to.)
Well that's a silly metric to use, it relies on people using Netflix for a starter. Akami still do their quarterly report and is most likely used by a lot more people (don't MS use their CDN for updates?) which will offer a better standard to judge average speeds over a greater population density. The UK's average speed isn't doing too badly, around 5.6Mb which is only going to get faster as BT continue to roll out FTTx (based on average connection speeds) :
Could be worse, Cuba just about manages bonded ISDN speeds.......
You perhaps underestimate how widely used Netflix is in the US?
There is no need to estimate, the article states one beeellion hours to thirty meeellion subscribers. I'm impressed netflix customers manage to watch more than an hour a day, I barely managed to watch three movies on my 6-week lovefilm trial. But then that's why I haven't signed up, innit.
"Average Speed for Netflix Streams"?? WTF does that mean?
I just checked, and I get 23 megabits on my Verizon Fios service and I don't have anything close to the top tier on offer. (I checked using an actual real world task, not a speedtest site - downloading the 80mb Linux kernel source tarball from kernel.org.) Netflix's numbers are off by an order of magnitude. I have no idea where these are coming from.
Maybe the stream only ever really tries upto about 3Mbps?
Rather, there is an OoM gap in the communication here. Netflix was reporting in megabytes so your 23Mb is close on to their 2.19MB. (well, not really close, you got 2.88MB. Not Bad at all, but probably well within their error bars)
Pirate, 'cause what do you do with the bandwidth when not watching movies?
They're actually reporting in megabits per second, not megabytes. The initial graphic that was provided, the one used in the story at The Register, was replaced on the official blog post with a less-ambiguous chart reading "Mbps", unstylized as compared to before.
That said, Netflix provides compressed video, so full utilization of bandwidth is not usually required. I use AT&T U-Verse and easily get my full advertised 12Mbps on most services and can usually use other services while someone is watching Netflix without any obvious performance loss.
Correct me if I'm wrong, but surely ~2.5MB/s (assuming that's bytes) is about as bandwidth-intensive as a video stream is going to get? That's 20Mbit, well in excess of what H.264 needs for good 1080p video. Pumping more data than that down the pipe would just be a waste. So, great for measuring relative rankings, certainly, but for absolute speeds? Pile of salt, please.
I'd just like to point out that 20Mbit (2.55MBps) is where Netflix throttles it's data connections to Google, not what the link is capable of. There's no point buffering half a movie that might not be watched on a well commissioned gigabit link.
Surely there's a hard limit to how fast a Netflix stream goes, there may be a bit of buffering at the start but at some point it is just what the video requires.
I'm assuming those top figures are about 20 Mbps (=2.5 MBps) which seems like a lot for an h.264 HD video stream. Whether someone has 20Mb or 200Mb is probably not going to show up in Netflix's figures.
People who subscribe to a lower cost lower bitrate plan would skew these numbers. I have a 20Mb/5Mb plan from 13th-ranked Centurylink, but where I live I'm eligible for 40Mb/20Mb. I just don't see the point in paying more for it, as what I have is fast enough for anything I do with it.
It would be more interesting to see how many people are ELIGIBLE for a certain speed of broadband, rather than who actually has it. Everyone with FIOS (maybe not all, but most) can get 300Mb/65Mb, but who really needs that enough to pay whatever Verizon charges for it? What real world benefit is there for a residential user to have 300Mb/65Mb versus 100Mb/65Mb, or the cheaper options, for that matter? Outside of being a P2P supernode, that is.
Exactly... surely the good people at SpeedTest.net are the ones with the best figures, which they collate into:
When a number of ISPs are giving you full untouched access to SpeedTest (and other speed testing sites), but then throttling/ downsizing/ cutting your capability everywhere else it starts to not become a very accurate guage anymore.
(FAIL for the ISPs doing this)
You guys are misreading sustained service levels across a site for a month, not bursts obtained through speedtest.net. The report is netflix bragging about how much data they can deliver per user, more than how connected the user is to their isp or to netflix's CDNs.
The Client that they use for streaming only uses about 2.5mbit at most for streaming of HD content, and depending on your connection speed it caches to a good degree then stops based on how many minutes ahead in the overall content you are.
What really needs to be explained clearly is the test metrics .
If 2.5mbit is really the max needed to maintain a stream than what they are saying is that when you reset a watched video the latency is so low that there is pretty much like sliding the silder back on your local media player when watching a local movie. Which is kind of impressive, But if this isn't the case its really not saying much besides that their client base is mostly on lower end connections.
Guys and gals, don't read the numbers as throughput. As others have said Netflix throttles their bandwidth per connection. Instead, look at it as a measure of network efficiency. Google is able to provide the 2.5 MB consistently better than the others which means users watching Netflix will see better quality video than those without.
I would kill for fiber to my home, but a 4MB connection costs me $200 a month.
"One of the main reasons for Google to get into the ISP business was to shake up the US market and get it back to the top of the world's broadband charts."
The only reason google got in the carrier space, isp/cell, was so that users didn't have an option to have their data collected and mined by google. Pffft, pretty easy to profile an advert target when you have ALL their data from the wire. Google has long passed do no evil.
30 mbps down, here.
"The only reason google got in the carrier space, isp/cell, was so that users didn't have an option to have their data collected and mined by google. Pffft, pretty easy to profile an advert target when you have ALL their data from the wire. Google has long passed do no evil."
Nope, it's because the asshats at AT&T wanted to double dip, they already charge their own customers for internet access, but thought that Google should ALSO pay AT&T because some amount of traffic was going between their customers and Google (especially Youtube.) Completely ignoring that Google already pays the standard fees at whatever peering points they are at, just as AT&T does. They were pushing congress to eliminate any semblance of net neutrality. So Google told them to piss off, and said if AT&T kept pursuing this Google would just start their own ISP, make sure to build out in all AT&T's wireline markets and take all their customers. After Google started building out, AT&T realized Google was serious and backed off.
As for these speeds.. the image incorrectly shows megabytes per scond, the one on netflix's blog now shows megabits. But, Netflix CAPS at 4mbps, so it's natural that these ISPs all show below 4mbps... it'll be a mix of 4mbps ratings and those times when the ISP is too crapped out to get that speed (or the user has other stuff going on on the connection). I.e. as they say the relative ratings should be useful (to tell which ISPs are either quite slow or overload a lot) but the absolute numbers are not real useful. The US broadband market is pretty bad (especially pricing), but no we don't order 20mbps service and get 2.2mbps or whatever.
Biting the hand that feeds IT © 1998–2017