You might know this but in case others don't, seeing different speed test results to different services/servers is completely normal.

A lot of people forget or don't understand the "net" part of "internet". The internet isn't a monolithic service you connect to. When your ISP offers you "1Gbps" internet, they're not guaranteeing that whatever your activity, you will get 1Gbps, they're just giving you a 1Gbps connection to their network.

Their network will then interconnect with other networks. Depending on the specifics of those interconnections, you will see different performance characteristics depending on what you're doing.

For example my ISP peers with a particular network that hosts servers ~600km away that I exchange a ton of data with. I get 1Gbps when I use my ISP. However when I use a different local ISP that doesn't peer with the remote network, I see significantly reduced speeds, because the route taken to exchange data isn't optimal.

I know the HN crowd is fairly technically competent but I've seen plenty of competent people who I think knew this in an abstract sense but it just hadn't clicked for them what the practical consequences were.

Thanks for the feedback, very much appreciated! I'm a product manager here at Cloudflare, responsible for launching this tool. Since the launch, we've found some issues that we're going to address:

- Especially for users with a very fast Internet connection, speed.cloudflare.com reports upload speeds much lower than expected figures. We don't yet know what is causing this but will disable the upload part of the test until we know more.

- In general reported download speeds are little lower than figures coming from other speed tests. We will revisit our methodology to understand the discrepancy.

- Re: the speed test automatically starting: we appreciate the feedback and understand why some users may not want this as default behavior. We will disable the auto-start for now.

In the meantime, we appreciate any and all feedback, please keep it coming: you can reach me at achiel [at] cloudflare.com

Another bug report: the pop-up info boxes are hidden off the side of the screen on mobile Firefox. Screenshot: https://cloudflare-ipfs.com/ipfs/QmRHQqSuJm4kMQB67hoiaYJ5fXc...

> - In general reported download speeds are little lower than figures coming from other speed tests. We will revisit our methodology to understand the discrepancy.

For what it's worth, Cloudflare shows me at 10mbps down, and Speedtest shows me at 160mbps (much closer to my expected 200mbps). This is a large difference.

Yes, that is. Want to drop me email (jgc @ cloudflare . com) with details?

speedtest.net (and I expect quite a few other speed tests) will open many connections to overcome slowness caused by TCP (low max window sizes and slow scaling). This is likely where the download discrepancy is coming from.

Personally I prefer the single TCP connection results as these tell me what I'm likely to see in a real-world situation such as web browsing with HTTP/2 or a large download.

I came here to make all the same feedback. My 1Gbps connection also massively under-reports.

Yeah. We're disabling the upload test until we figure out why.

I'm also getting 10 - 43mbps and 75ms+ ping w/ 60ms jitter on speed.Cloudflare (vs 550mbps and <10ms on fast.com).

I've confirmed the 500mbps speed I pay for in Montreal is accurate with my own iperf and iperf3 tests to physical servers I own in NYC, so it's not a "your ISP is colluding with the speedtest sites" thing. I've also confirmed I have a 1ms rtt to 8.8.8.8, and a 10ms rtt to 1.1.1.1 with ~2ms jitter by pinging them in terminal.

These CF test results worry me somewhat because I host a bunch of traffic on servers in my closet over Cloudflare Argo tunnels, does this mean those services are only able to push ~45mbps with 60ms ping via Cloudflare? Or is this just an artifact of something weird going on with the test methodology?

What do the little “i” icons do? I can’t click them using Safari or Chrome on an iPad.

Hovering it opens a tooltip with details

It seems to still be auto starting on page load for me. Chrome Android

Don't let Viasat cheat your system...

For Bitcoin, we run a open source relay network called FIBRE, and incredibly low latency network which uses some encoding tricks to gain significantly faster than TCP transmission of block data around the world. It's currently limited in part by the number of points of presence that can be reasonably operated from. Is there anybody at Cloudflare that would be interested in taking this on using your incredibly latency-diverse network?

Note: ISPs provide speedtest.net servers themselves, so that never leaves the ISP and only tests the link to them.

fast.com likely has the exact same issue, as Netflix has content boxes at ISPs, although I can't say for sure.

Plus, even then, "speed" isn't absolute. It's all about agreements, routes, capacities and load. Test to specific targets of interest if you can, e.g. http://speedtest-nyc1.digitalocean.com/.

It depends what you want to measure of course. I provided results to a variety of servers besides my local fast.com and speedtest.net server. As well, I'm familiar with what speeds I get when downloading Xcode, when pulling binaries via Usenet, etc. The results from Cloudflare indicate to me a bottleneck on their end, and something is especially wonky with the upload result.

I get 860-910 Mbps down/ 180-372 Mbps up to NYC[1-3], TOR1, but I have to test using Chrome. Running the DO test under Safari is pegging the CPU on a Macbook Pro 2.7 Ghz i7. (The other speed tests run fine under Safari.)

Indeed, their tests are currently wonky. I get 10% of my regular upload capacity, and 30-50% of my download through their test, despite testing to a Cloudflare location within the same city as myself.

I just wanted to ensure that we correctly discredited useless numbers and focused on the interesting ones instead.

Doesn't cloudflare also have boxes in ISPs? Youtube, Netflix, Facebook, Akamai, Cloudfront, etc all do. It's all relative.

Do they? In that case, this metric is also pretty useless. Unless your only interest is an internal network with your ISP.

Cloudflare is reporting the p90 throughput of samples. Doubt the others are measuring this way.

Even the max rates reported under "Download Measurements" and "Upload Measurements" are slower than any other site I test against. The site also has a lot more variability from test to test.

I get download measurements anywhere from 400-800 Mbps, but mostly under 500 Mbps. I have yet to see an upload measurement above 100 Mbps.

This test just doesn't reflect real-world results for my Internet connection.

Not to discount anything you've said, but some speed tests I've run find a server that is so close to me that it's not really testing 'Internet speed' as it is just the speed of my ISP and maybe another. I don't know what people want from speed tests and it's good to have different ones, just know what you're measuring.

I'm not a low-latency gamer so none of the differences really matter unless they're near zero. I've been streaming video and fetching/pushing git repos that I use since having a 5Mbps connection.

Interestingly, for me, it's reporting significantly higher speeds than I'm paying for, and than other speed tests report.

I'm paying for 100 Mbps down / 100 Mbps up, and every speedtest always comes very close. Upload usually a little higher. From experience the actual maximum speed is also roughly 12 to 13 MBps.

Cloudflare is reporting 667 Mbps down / 220 Mbps up.

It depends how it's tested. Clearly short burst test produces high numbers, because it gets really high, before traffic shaping (throttling) kicks in. In my case cloudflare reports almost 10x the speed I'm actually getting for prolonged downloads. Of course burst speed is much higher.

It's also showing higher speeds, for me. Though only ~20Mbps higher on Cloudfare's speed test vs. other speed tests (and what I'm paying for).

Isn't this because Netflix (fast.com) has better peering agreements with most ISPs while Cloudflare relies more on their own infrastructure?

Don't think you can compare those two, as they are different things. Even if Cloudflare "relies more on their own infrastructure", they still have to have peering agreements with others, otherwise they can't accept/send traffic.

More likely, Netflix has better infrastructure and better peering agreements, than Cloudflare. Which is kind of surprising, since Netflix is supposedly a media company and Cloudflare is a "internet" company.

Netflix has an advantage: their OCA boxes may handle speed test traffic [1]. That means a box racked at your ISP[2] may be serving that traffic. While Cloudflare may have nice peering agreements, they don't have that.

[1]: https://netflixtechblog.com/building-fast-com-4857fe0f8adb

[2]: https://openconnect.netflix.com/en/#what-is-open-connect

How does that explain me getting solid results from DSL Reports, Ookla to a random server I picked in Ashburn, and to Google Fiber's servers in Atlanta, Charlotte, and locally.

I think there's something about the methodology of Cloudflare's test.

My point is limited to "fast.com could give faster results because you may not even leave your ISP to talk to them"; I'm not here to say "Cloudflare's results are good/bad/better/worse/reliable/unreliable."

A thought, though: Cloudflare reports your p90 time as "your speed". I don't know what the other sites report. Is it the same?

Or the ISPs don't throttle traffic to fast.com

That makes sense, think there been evidence of ISPs throttling connections, except for speedtests, in the past, so would make sense. My own ISP also used to throttle Steam downloads for a couple of years, until suddenly they didn't.

I thought one of the goals of fast.com was to let users find out if their ISP is throttling Netflix. ISPs can't only throttle a speedtest if Netflix makes it indistinguishable.

Maybe that is, but one can think of a number of ways for ISPs to work around that. Quick thought: if the ISP is also your DNS resolver (which is the default for most people), check if there been a query for fast.com and if so, stop the throttling for five minutes. After those minutes, start throttling again.

Not so, Viasat will throttle based on headers and even built helpers for speed tests that bypass other limits. Such as forced video resolution down grades.

I thought Cloudflare also peered caching servers?

Meanwhile in a major city in Australia:

44 Mbps down 30ms latency 12.5 ms jitter

What’s it like having internet that actually goes?

Luxury! 33.4 Mbps down 22.5 ms latency 8.5 ms jitter in a medium city in Tennessee, USA. Paying for up to 100 Mbps, but ~35 is consistent across times and sites.

Follow-up: 33.4 was because I'm an idiot. I have a surge suppressor with coax in/out between the wall and my modem (lost a cable modem, router and networked laser printer to lightning a few years ago). Removing it puts me at 103/23.0/4.53. I'm sure I checked after adding the surge suppressor, but not, apparently, in quite a long time.

Meanwhile in a minor town in Australia:

26.9 Mbps down 11.4ms latency 1.96ms jitter

What’s it like having internet that actually goes?

(my synch speed is only 24.73 Mbps, so those numbers dont seem right...)

Come on guys, 25mbps is enough for like 4 SD Foxtel streams. That's enough for anyone.

-- Tony Abbott, probably.

Metropolitan Sydney (upload is temporarily disabled)

Paying A$89 for TPG 100Mbps plan

Server location: Sydney Down 93.6Mbps Latency: 14.9ms Jitter: 4.97ms

Before students went back to school, upload bandwidth was throttled at 4~8Mbps (due to the congestion caused by nation wide WFH), now speedtest / fast.com reports 30~40Mbps...

Surprisingly, download bandwidth has not been affected like Optus Cable before (experienced 10Mbps or less for 100Mbps/2Mbps plan), facepalm.

I strongly prefer google's speedtest[0] as it's run through M-Labs (Measurement Labs). This group originally gained prominence doing distributed testing of ISP's to determine if they were blocking specific types of traffic (torrents, etc) I think 15 years ago or so.

M-Labs for me has the highest real-world accuracy, because it's designed to obfuscate the testing servers so that your ISP can't cheat or make itself look better by prioritizing M-Labs servers.

0: https://www.google.com/search?q=speedtest

Netflix has servers directly in the ISP datacenter.

Ookla also is generally run out of the isp datacenter.

Cloudflare has servers running at a PoP and doesn’t get the same performance

For me I get roughly the same download speed as the normal measurements (875Mbps), but upload is more than halved (423Mbps) compared to what other tests report (tests that I have more faith in, like bredbandskollen.se which is reputed to be very good for Swedish connections).

I also have synchronous gigabit fiber, and I was surprised to get way different results from the others. They're similar to yours. Not sure if this is on CloudFlare's end, or if there's something on my end or my ISP's end.

Seeing the same so most likely on cloudflares side.

These are just like the results I get for 1Gbit comcast here in the bay area for the various services.

Now you're just showing off ;)

No I think this is important. Are they that wildly different from each other that they aren't that useful? Is one better than the rest?