Netflix Launches Fast.com To Show How Fast Your Internet Connection Really Is (venturebeat.com)

Paul Sawers, writing for VentureBeat (condensed): Netflix really wants to show you how fast (or slow) your Internet connection is, and to do so it has launched a new website at Fast.com that conveys the real-time speed of your connection to the Web. It's designed to give people "greater insight and control of their Internet service." Netflix said it was for: Providing a website featuring non-downloadable software for testing and analyzing the speed of a user's Internet connection, as well as downloadable computer software for testing and analyzing the speed of a user's Internet connection.Compared to Speedtest.net, Fast.com doesn't offer any details on how fast is your upload speeds, what's the ping time, and any detail on location and ISP. However, it's seemingly faster, and automatically detects your download speeds when you visit the website.

Comments

Works pretty well

[Anonymous Coward]

Got my connections DL speed correct in about 3 seconds.

Re:Works pretty well

[phishybongwaters]

Your burst speed means nothing. I was going well beyond my DSL cap if I only cared about burst speed, the EFFECTIVE speed is considerably lower. Any speed test site is only as good as the infrastructure, and once it's popular, ISPs will unthrottle connections to that service, much like they have agreed to do with netflix, and it becomes speedtest.net version 2. If you want to test your speed to the internet, reliably, you have 1 choice. Host a webserver yourself somewhere, and upload files to it. That's literally the only test you can trust.

Re:Works pretty well

[mindwhip]

Their claim that they are giving CDN representative speeds but they aren't doing this properly.

It gives me what my typical US>UK transfer rate is (70MBs) for most speedtest nodes in the USA so I'm guessing that's where their server is testing from.
However my typical speed across most of Europe will get near to my ISP rated speed of 140MBs with speedtest and I have seen game downloads etc actually hit this speed so this is only telling me what my speed to the US is and not what my *actual* speed to most CDNs is.

Re:Works pretty well

[flargleblarg]

It gives me what my typical US>UK transfer rate is (70MBs) for most speedtest nodes in the USA [...]

You're getting 70 Megabytes per second???

Re:Works pretty well

[war4peace]

I'm getting that too, and sometimes above it, provided my HDD can keep up.
Downloading to SSD reaches the maximum theoretical throughput of an Gigabit connection.
And all that in what's perceived as being a third world country, where my Gigabit Internet, bundled with basic TV (65 channels) and a free uncapped 3G dongle costs 20 bucks a month.

Re:Works pretty well

[mindwhip]

oops that should be 70Mb/s to the states and 140 Mb/s across Europe... I blame my crappy phone touch keyboard and autocorrect for caps issues :)

But i've got FTTC Cable and there is way more capacity there if my ISP felt they could/should offer it...

My ISP offers upto 200Mb/s (and are rumoured to be beta testing 350Mb/s) but I don't see the point as I'm already past the point where it makes much noticeable difference for anything...

Re:Works pretty well

[Anonymous Coward]

I got 180Mbs via https://www.voipreview.org/speedtest
and 171Mbs via https://fast.com/

from Xfinity here in Massachusetts

Re:Works pretty well

[tattood]

The 'whoosh' here is that he's being pedantic about the fact that data transfers are usually measured in megabits per second (Mbps) not megabytes per second (MBps)

Re:Works pretty well

[citylivin]

"That's literally the only test you can trust."

i find speedof.me to be very reliable. It tests differently then the other tests, trying to simulate actual traffic.

http://speedof.me/

Re:Works pretty well

[Bengie]

It is a good tool. No one tool is perfect, especially with the Internet where any metric can be gamed in some fashion.

Re:Works pretty well

[I4ko]

try speedof.me

Re:Works pretty well

[pepsikid]

Eh, no. I have Google Fiber. I've performed about 10 tests in a row now. First, it measured my speed in the mid 20 and mid 50 Mbps a few times, and now it's showing my speed as a nearly rock-solid 100Mbps for most subsequent tests. There should be more variation, and It should be showing 4-10x faster speeds (based on speedtest.net and google's speed tester). This thing isn't accurate.

Re:Works pretty well

[Bengie]

Last night around 8pm, I ran the test 30 times back-to-back, and each time I got between 90 and 92, most of the time, 91. Mind you, my 100Mb network is also doing other stuff. But this morning, I ran it and got only 73. Not expected to get a worse value at 7am.

Re:Works pretty well

[Beezlebub33]

Also, download speeds are mostly worthless. Upload speeds are where the money is.

Um....what? Not for home use, it's not. For businesses, upload speeds are important, but for people streaming content to their homes (i.e. the vast majority of consumers), download speeds (and consistency) are what they care about. That said, you are correct that the speedtest sites are a poor test since ISPs shape their traffic to do well on them.

Re:Works pretty well

[pepsikid]

To clarify, I'm saying that fast.com keeps saying that my Internet speed is *exactly* 100Mbps. And I have a 1,000Mbps connection anyway.
Fast.com won't report my connection as any higher than 100Mbps but occasionally reports that it's slower. I checked my settings and performed some tests; my NIC is at gigabit speed. My switch is showing a gigabit connection, I can xfer files between PCs on my lan well above 700Mbps and most other (gigabit capable) Internet speed tests show I have 300-700Mbps.

Fast.com fails!

How long before the ISPs...

[Anonymous Coward]

...start to provide better speed to all requests for just this site and game the results?

Re:How long before the ISPs...

[ninthbit]

Even if its on the same ports on the same servers and using encryption, the ISP will just need to set their throttling to kick in after a short delay. Then the speed tests will complete quickly, and the longer streaming video is still limited. If you really want to test your speed, you pretty much need a private VPN with something like a Linode instance and openvpn. Of course then you have to account for the overhead of encryption and an extra hops; but at least you can rule out ISP shaping.

Re:How long before the ISPs...

[swb]

I've always wondered to what extent bandwidth measurement sites worked to defeat ISPs gaming their sites.

When I would do this the old fashioned way with FTP, I would always use a test file built from garbage from /dev/random, run through DES a couple of times, and then run through gzip, trimmed to the size file I was willing work with. I wanted to make sure there was no possible redundancy in the data anywhere.

Do measurement sites even do this with their measurement data?

You would think they would want to download data that can be cryptographically verified n the client end to prevent ISPs from delivering dummy packets or forging the session to improve throughput and hide bottlenecks.

Ideally Netflix would be just using their own infrastructure in place to fetch a random chunk of video from their servers via the netflix.com domain name, so that from a network perspective it looked as much like a movie stream as possible. Maybe interleave non-video data with it from a separate source to validate that it isn't being affected by netflix-specific shaping.

They could go further down the rabbit hole and spin up short-lived AWS instances with AWS dynamically assigned IPs to provide the server end of testing to further thwart throttling or gaming of traffic.

Re:How long before the ISPs...

[Gr8Apes]

3s isn't nearly long enough. I know that U-verse, for instance, doesn't kick in throttling your connection down to the "maximum" for about 5s, so if you have a largish file that can be squeezed into a 4s burst, your file gets transferred in 4s. Double the size, and it may take 30s. Yes, the differences are that dramatic.

Works great

[U2xhc2hkb3QgU3Vja3M]

No Flash, no Silverlight, got my cable download speed accurately.

Re:Works great

[whipslash]

Here's another good one that doesn't require Flash or Silverlight but gives a lot more details: https://www.voipreview.org/spe...

Re: Works great

[Anonymous Coward]

> have you actually called up the cable/DSL/fiber provider

Seattle has two government-protected cable monopolies that divide the city. I live in the Comcast area, and they have no incentive to ever offer service since no competitor can. Digging up streets and replacing pedestals is expensive so they don't do it. Also, the city fights upgrades and requires a super majority of affirmative votes in order for them to be allowed to install or upgrade service. A nonvote, like from a home under foreclosure or a rental unit where the owner isn't available, counts as a no vote. It's nearly impossible to get permission to even add a pedestal. I know. I've been fighting to get service to my street for over a decade.

In addition, the phone wiring is very old in most of the city and there aren't many central offices so DSL simply isn't available to much of the city.

Translation:

[Sir_Eptishous]

NF: "See, we TOOOLLLDDD you it was your ISP!

Selective throttling

[JustNiz]

I've long held a theory that my ISP (Cox) is limiting bandwidth selectively by site, and that they make all the benchmark sites wide open, but throttle others, like netflix.
It seems they have already added fast.com to their "Do not throttle" list but not added Netflix.

Re:Selective throttling

[wile_e_wonka]

You've got it backward. They haven't "already added fast.com" to the whitelist, they just haven't added it to the "throttle list." Netflix gets throttled, bittorrent gets throttled, speed test sites and most other sites don't get put on the "throttle list."

That said, I have Cox on the 100 MB/sec down plan and have not had any throttling issues, even when well above their soft data cap.

Re:Selective throttling

[JustNiz]

I don't see how not using their DNS changes anything.
I'll bet they are resoliving the selective throttling to IP addresses at some level prior to the actual decision point.

Re:Selective throttling

[kwiecmmm]

I have the same ISP (Cox) and I have never had any issues with their Netflix bandwidth speeds. Previously I had Charter, and would occasionally see issues there, but nothing like what Verizon did to Netflix.

Re:Selective throttling

[known_coward_69]

a lot of media streaming CDN's use DNS resolving to find the closest server to you. if you're on the east coast and use some DNS server 2000 miles away then that's where your content is going to be streamed from. not a local server

Re:Selective throttling

[Solandri]

The new fast.com site test seems too short. Cox will give you more bandwidth than you're paying for for a few seconds (I'm on Cox) - to speed up bursty use like general web browsing, but slow down large downloads. You can see this on the speedtest.net tests. I'm on a 100 Mbps plan and it initially starts at around 160 Mbps, then gradually decreases, usually finishing around 110 Mbps. So fast.com is reporting your burst speed, not your sustained speed.

I've started using internethealthtest.org more. It runs separate tests to major backbone providers. Usually one or two are slower (often a lot slower) than your connection speed, indicating a bottleneck on the route your ISP uses to their networks. This is often the cause of some sites being fast while others are slow. You can also compare different ISPs in your area using MLab's Observatory tool.

Re:Selective throttling

[Sir_Eptishous]

Throttling by individual IP...at an ISP level is hilarious.

Its like whitelisting all the MS Office 365 / Exchange Online ips

Re:Selective throttling

[Anonymous Coward]

A little bit of digging through Fast.com's traffic and it appears that the traffic test is against Netflix's CDNs. This means that the IP address is the same for the video as well as for the speedtest. Also the fast.com traffic to the CDN is over SSL which makes it very difficult to distinguish between test data and movie data.

I suspect that it would be very difficult for ISPs to distinguish between the fast.com's speed test and actual movie traffic. And simple whitelisting isn't going to cut it.

Re:Selective throttling

[BitZtream]

1Gigabit over DOCSIS isn't even a little difficult and is done in several markets already.

5 or 6 gigabit on current DOCSIS 3 hardware is unlikely to work, 1? No problem. 1gigabit == 125 megabyte

Re:Selective throttling

[Etcetera]

I've long held a theory that my ISP (Cox) is limiting bandwidth selectively by site, and that they make all the benchmark sites wide open, but throttle others, like netflix.
It seems they have already added fast.com to their "Do not throttle" list but not added Netflix.

So... do you have any evidence for your theory? Or is this just a theory? Outside of emergency/defensive purposes, ISPs only "care" to the extent that there's a financial reason to care and, depending on the site, it makes far more sense to host a local caching server for a high-volume service (eg, Netflix) than throttle it in a way that anyone will notice.

This all depends on ISP size and connectivity, of course, but FWIW I've had Cox for 15 years in San Diego, worked at ISPs here in the city from 2000-2008, and have never noticed any particular limiting that wasn't most likely -- and easily explained by -- technical issues.

Re: Selective throttling

[JustNiz]

Link please? then I can get actual metrics to backup my theory

Re:Selective throttling

[Etcetera]

a lot of media streaming CDN's use DNS resolving to find the closest server to you. if you're on the east coast and use some DNS server 2000 miles away then that's where your content is going to be streamed from. not a local server

^This.

I can't believe how many people don't realize this, put in Google's public DNS because they think they're being cool (and not being tracked), and then complain that an A record comes back that's 18 hops away from them.

Re:Selective throttling

[Etcetera]

That said, I have Cox on the 100 MB/sec down plan...

I kind of doubt you're on the 100 MegaByte plan. It's doubtful Cox could even deliver those speeds.

He probably means Mbps, but FWIW Cox does have a 1Gb plan in some markets.

Re: Selective throttling

[nazrhyn]

https://www.netflix.com/title/...

Re:Selective throttling

[omnichad]

I'm pretty sure Google's Public DNS uses anycast IP addresses. Meaning you'll probably still get the closest server in the results. A lot of CDNs do the same thing.

Re:Selective throttling

[guruevi]

The problem is that TWC for example returns addresses to servers they colocate for Google and Netflix but then either rate limit or overload those colocation centers so they don't have to peer.

Re:Selective throttling

[unrtst]

When you use their DNS, they know[and can throttle] your connection by site[domain name].

That's not how things work. DNS may be used to direct you to a closer/faster IP, which is essentially how CDN's work, but they're not using it to determine if they should throttle the connection or not. DNS is disconnected from that equation. Your device makes a lookup for the hostname, gets the result, terminates connectivity with the DNS server, and then establishes a connection to the IP(s) that you found.

Otherwise they'd have to do reverse lookups which would make "site" throttling a bitch.

I'm uncertain how they maintain their lists, but using reverse lookups would be trivial. You don't have to do it on every hit - just keep a local cache. Anyone that's ever done web server log analysis, or any log analysis really, already knows and does this.

Re: Selective throttling

[CrankyFool]

Fun fact, BTW: That movie was filmed on the Netflix campus

(But I still have no idea who that guy is. Pretty sure he's not a regular Netflix employee)

Re:Selective throttling

[Bengie]

From what I've read, gigabit-magnitude speeds. Other than the first few people who were the only ones using it, other are now showing they may get the infamous "up-to" 1Gb, but actually get anywhere from 600Mb/s to 1Gb. And of course they still have issues trying to buffer a 5Mb/s Netflix or Youtube video.

Those speeds are assuming you're using a local speedtest server. Once you're outside of their network, you may be getting closer to 200Mb. Then you see people with Google Fiber testing their connections all around the USA and parts of the world and still getting 900Mb+/s.

Re:Selective throttling

[Bengie]

internethealthtest.org doesn't work for me in any of my browsers. It always gives me 70Mb down and 20Mb all the time to all of the servers. At least it's consistent to within 0.5Mb/s on every run. I have 100/100.

Better watch it.

[will_die]

Some company is using this to collect information about your computer and network and then attempt to sell you something based on the information they collect.

Re:Better watch it.

[funwithBSD]

Then don't buy anything.
Problem solved.

Seriously, I would never buy from a company that came to me, that goes for online or offline. I do my research, find the product that is best fit and price I can afford.

Re:Better watch it.

[BradleyUffner]

The horror

Re:Better watch it.

[MachineShedFred]

They're already selling me that thing, and giving me a tool to tell my ISP that they are interfering with that sale.

Netflix wins, I win, the ISP loses through their customer base having another data point to how they are being screwed. I see no problem with this.

Re:Better watch it.

[Sir_Eptishous]

Netflix wins, I win, the ISP loses through their customer base having another data point to how they are being screwed. I see no problem with this.

How do you win when you can't get your ISP to change who/what they throttle and there is no competition for ISPs?

downloadable computer software...

[Jonah Hex]

Downloadable computer software for testing and analyzing the speed of a user’s Internet connection.
There is no word on any downloadable software yet, but the website is certainly now live and it’s a fairly straightforward offering.

Anyone find the downloadable version, maybe a beta? It's not mentioned in their weirdly styled "?" help section.

Re:until it gets popular

[grahamsz]

Which for netflix is actually entirely possible. It'll also draw more attention to ISP shenanigans if fast.com doesn't agree with speedtest.net

Slashdot becoming irrelevant?

[rjejr]

The site hasn't crashed yet, what happened to the /. effect? It gave me 56, the Speedtest that it linked to gave me 57, but that one took much longer, I'm getting 50, so I think for a really quick check of your speed it's not a bad thing. I'm going to check again around 9PM, things always bog down between 8 and 10.. Just ran it again, still came back with 56, so it's consistent.

Re:Slashdot becoming irrelevant?

[mspohr]

I got 47 from fast.com and 65 from speedtest.net. Charter cable.
(Tried it several times and still got a significant discrepancy.)

Re:Slashdot becoming irrelevant?

[sims 2]

I'll blame this one on netflix being used to handling several million people streaming video at 5 Mbps.

On at&t u-verse ADSL2+ it shows 11 Mbps that's pretty close considering it's on a 12 mbps plan.

Re:Slashdot becoming irrelevant?

[ShaunC]

The Slashdot effect doesn't work so well on sites deployed to Akamai.

Re:Slashdot becoming irrelevant?

[mdm-adph]

As we all get older, there's less and less people adding into the "slashdot effect" yearly...

Re:Slashdot becoming irrelevant?

[green1]

25 meg plan here (TELUS) showed 27 meg which is what I get when downloading files and matches speedtest.net, unfortunately when I ran it a second time it showed 6 meg, and now refuses to go higher, but speedtest.net and actual downloads from various sites still show 27 meg.
So something isn't right here.

Will need to check during prime time

[Ogive17]

I got 30+ mbps but everyone's at work or school right now. I'll be curious what it shows at 8pm tonight.

Never stop never stopping

[pr0t0]

Could not reach our servers to perform the test. You may not be connected to the internet

If I wasn't connected to the internet, I wouldn't see the page indicating I may not be connected to the internet.

Re:Never stop never stopping

[Anonymous Coward]

Could not reach our servers to perform the test. You may not be connected to the internet

If I wasn't connected to the internet, I wouldn't see the page indicating I may not be connected to the internet.

Keyboard error. Press F1 to continue.

Re:Never stop never stopping

[yes-but-no]

Rendering that html may trigger an exploit which fries the RJ45

Re:Never stop never stopping

[Ulric]

Could not reach our servers to perform the test. You may not be connected to the internet

Abort, Retry, Fail?

Suspiciously big differences? Or not?

[wjcofkc]

So fast.com says my download speed through T-Mobile is 14 megabits. Speedtest.net say 60 megabits while speedof.me gives me 5.7 Mbps. This is using my phone as a hotspot. Hopping all around the internet and randomly downloading very large files including the .iso files for Ubuntu and Slackware and with the case of Slackware I hit up multiple servers around the world I can safely infer the 60 Mbps is a close approximation as an average. Don't get me wrong, I know how the internet works, but still. Oh well, I have long advised the ranking on speed testing sites and apps are most likely bought and paid or at the very least biased in someway. Proof? Dunno.

If your wondering, my business account with T-Mobile affords me approximately one metric fuck ton of data per month.

Re:Suspiciously big differences? Or not?

[Ingenium13]

Do you have binge on enabled? If so, that will throttle Netflix connections. It's enabled by default on all accounts unless you go in and disable it.

Re:Suspiciously big differences? Or not?

[wjcofkc]

Fuck I forgot about that.

Re:Never

[sims 2]

The att u-verse business line here is $45/mo for 12 Mbps but that's with a discount since we also have phone service it would be $50/mo by itself.

The deal is that the city is offering a 10/10 mbps business fiber line for $55/mo and prices have been adjusted appropriately.

Who needs a test when you have real-life app data?

[Cajun Hell]

If this is primarily intended for Netflix users, then why not just have the client measure the speed of actual live use cases? Some games have an option to show frames per second; your streaming player could have an option that shows bits per second.

VPN Users Affected

[ZeroNullVoid]

With this using Netflix's servers and their VPN blocking, I get the following error when testing via various VPNs.

* Could not reach our servers to perform the test. You may not be connected to the internet

Mod parent up

[Solandri]

Oooh, that's clever. Set up a speed test site which if the ISP prioritizes to speed up their results, also speeds up Netflix's CDNs.

Not even close.

[wolrahnaes]

My connection is nominally 250/25.

Speedtest.net gives me 238/28 to another ISP across the state from me (http://www.speedtest.net/result/5335405259.png). Amusingly I actually get a bit worse, 220/28, to my ISP's own Speedtest server (http://www.speedtest.net/result/5335408660.png)

My usenet and Steam downloads agree, I can easily max out my connection with either.

Fast.com gets me between 35 and 45 Mbit/sec down.

Re:Not even close.

[Ingenium13]

Then your ISP is likely throttling netflix connections, or there is a bottleneck between your ISP and Netflix. That is the entire point of this site. Some ISPs throttle Netflix connections. This is to find out if yours does. Mine (Comcast, SF Bay area) tests at 170 mbit. Speedtest.net is the same. I pay for 150/10.

Re:Not even close.

[wolrahnaes]

Throttling Netflix to a number that's well over any rate they actually stream at doesn't really make sense. Netflix says 25mbit/sec for 4K and 5mbit/sec for 1080p, so if it's actually being throttled the only way that would ever affect me is if I tried to stream two 4K movies at the same time. I'm not even sure if that's allowed by Netflix on a normal plan, I'm pretty sure I pay extra to be able to stream two things at once even in 1080p.

Re:Never

[sims 2]

Have you checked around to see if any other dialup providers service your area? last I checked netzero was still selling service nationwide for $9.95/mo and still allows for 10 hours a month free/w ads.

Re:I don't think so

[bruce_the_loon]

Browsers and Bittorrent clients report download speeds in kilobytes or megabytes per second, this site reports download speed in megabits per second. 1 megabyte per second is around 8 to 9 megabits per second given overheads. Your 5 megabit/s line will reflect in the browser as 600 kilobytes per second, so the site is confirming your experiences.

Maybe not

[goombah99]

Fast.com reported > 30% faster speed than speedtest.net for me (18 versus 22) so I don't think they are sandbagging this. On the other hand it's not prime movie watch time of day either.

But for me, amazon is able to stream shows in HD better than Netflix is. My neflix connection rarely acheives HD quality even though I have it set to request that. Since this is on all my idevices it's not a matter of the computer or client software.

So either netflix has shitty servers or they get throttled at some peering level.

I will be very interested to see how this tool works when I'm experiencing a bad netflix movie stream. If it can help them prove throttling I'm all for it.

Re:Maybe not

[pnutjam]

Every evening around 8 - 10pm, my internet becomes unstable. HBO go, netflix, even when i was trying to order something from Walmart...
comcast...

Re:Maybe not

[Bengie]

When I watch Netflix at 8-10pm, they cram 1Gb/s down my pipe, but that's the most I can measure with my 1Gb Ethernet port. And no, my ISP does not have Open Connect, all over transit.

Re:Maybe not

[StikyPad]

Interesting... Speedtest.net consistently reports a speed ~20% faster than my subscribed Comcast tier (180Mbit actual vs 150Mbit advertised). I'd always assumed they just prioritize traffic to speedtest et. al., and that they'd just add fast.com to the "fast lane" now. It would be easy to test by assigning a secondary IP and domain name to a speedtest server and comparing results.

Download are done from nflxvideo.ne

[JcMorin]

I did a quick lookup and the download are made from: https://ipv4_1-cxl0-c273.1.nyc....... I believe this is the same domain as the real video so that would make it harder to block on "fast.com" since data are not downloaded from that domain name.

slow.com

[sims 2]

Since we are discussing fast.com does anyone know the story behind slow.com it says Welcome to Comcast! in google search results but it says Welcome to Time Warner Cable! if you visit it.

Did they set that up or was it someone's idea of a joke to redirect that domain?

Re:Who needs a test when you have real-life app da

[Nkwe]

Up, Up, Down, Down, Left, Right, Left, Right, Up, Up, Up, Up. YMMV.

Re:Never

[pnutjam]

10/10 beats the pants off 12/3 (more like 12/1.5) from Uverse.

Worthless!

[darkain]

I'm on Gigabit FTTH. Speedtest.net gives me anywhere between 500mbps up/down to 850mbps depending on time of day and testing server selected. Fast.com is consistently giving me 20-25mbps results.

Oh also as a note: fast.com at least from my location, is resolving to Akamai Technologies servers. It is also resolving to a server in San Jose when I'm in Seattle. So the link speed anywhere in between could explain the extremely slow connection compared to the local Seattle based servers that Speedtest.net gives me.

Bad Methology?

[Rashkae]

The speed reported on fast.com is less than 30% what I'm recording on my own network monitor while running the test.

Re:Bad Methology?

[Rashkae]

Arrrgh, what I meant to say, is that the speed reported is 30% Lower than what I measure. *sigh*

Messed up

[Theovon]

That link to the article is seriously messed up. I canâ(TM)t scroll down, because whenever I try to, something causes it to jump back to the top of the page.

Re:Never

[sims 2]

ADSL2+ so its actually 12/1 Mbps instead of the 12/1.5 Mbps that is also sold as uverse in most places.

Will have to switch providers sometime soon 1Mbps upload is quite limiting in what you can do.

Comment

[WallyL]

55 Mbps, which is close enough to what I pay Charter for (60Mbps). YMMV

Dishonesty, sneakiness, and other abuse is common

[Futurepower(R)]

"... once it's popular, ISPs will unthrottle connections to that service..."

That's my experience. SpeedTest shows users what the internet providers want them to see.

Businesses in the U.S. amaze me. Dishonesty, sneakiness, and other abuse of customers has become common. A HUGE example: Microsoft Adding More Ads To Windows 10 Start Menu.

speedof.me

[blackomegax]

I've been using speedof.me (flashless, javaless) as it presents a real time graph of all variables.

A bit overloaded?

[guruevi]

My connection says 75Mbps even though a Speedtest says ~200Mbps and another test says ~400Mbps (this is a business line) so I think they may be a bit overloaded.

Only for Netflix

[GuB-42]

This tests how fast your connection is with Netflix servers. Not your "real" speed if there is such a thing.
I have gibabit fiber, with Speedtest.net I get 900+ Mbps, which is my real speed when there is no bottleneck.

In reality it depends on what's on the other side. Fast.com tells me I have 320 Mbps, YouTube gives me less and Steam gives me more. They should call it the Netflix speed test instead of presenting it like some universal truth.

Re:way off

[green1]

For me fast.com was dead on for the first run. Matched perfectly with speedtest.net and every download that I do. However when I ran it a second time it gave me a result less than a quarter of that speed, and now won't go back up to my actual speed.
I'm betting a lot of ISPs are about to be flooded by calls from users who don't understand the internet complaining that this new site says their connection is slow even though the ISP is actually providing the speed the customer paid for.

Re:none of these are accurate

[hambone142]

I live in a rural area and only have access to satellite internet and one WISP. The WISP gives me a 2.2 Mbps reading but if I download anything for 10 or 15,seconds of more I get throttled to 120 Mbps.

Re:none of these are accurate

[hambone142]

Ack. That should have been 120 kbps.

Re:Not enough data

[Bengie]

speedtest.net can lie. Using PFSense, I've seen my raw WAN incoming traffic peaked at 30Mb/s with 1 sec averages, but was around 25Mb/s for most, and speedtest.net claimed 45Mb. I've tested this many times over the past few years. It really depends on the test server, and mostly affects slow servers with short tests.

My ISP has its own speedtest server and it takes a good 30 seconds to finish the download on my 100Mb connection.

Re:way off

[Bengie]

I was running one test after the other and kept my connection 100% saturated got almost 6 minutes. 100Mb, personally shaped to 99Mb/s via PFSense, averaged 98.55Mb/s per minute for 5 minutes and around 70Mb/s for a fractional minute when I stopped re-running the test. Seems stable to me. Every test is +-1 Mb.

Netflix should stop banning VPNs

[allo]

People have a right of privacy.

fake

[Robzoro]

my net plain is 1Mbps but this site showing me that my current speed is 3.32Mbps. lol i don't think its working for me

who should i believe?

[xuvetyn]

fast.com or speedtest.net? there's quite a discrepancy. =/

Comparing business now with business then.

[Futurepower(R)]

Note that when I said, "Businesses in the U.S. amaze me", I was intending to compare businesses now with businesses in former decades, when it seems to me there were many companies that were community-minded.

Download speeds only, serfs

[rcharbon]

Note how the test only shows download speed, because our corporate masters want us to be good little consumers, not competitors/