Making home Internet faster has little to do with “speed” | Hacker News Reader

Making home Internet faster has little to do with “speed”

blog.cloudflare.com

33 pointslladnar19 comments

https://web.archive.org/web/20230418160459/https://blog.clou...

Question to smarter than me (even when I work in telco these days) - does subscriber ratio matter for international bandwidth? Typical connection here is 1gbps, but you can get 6gbps. Assuming same 1:30 subscriber ratio that’s still a 6x improvement on international speeds (which matter so much when you are outside us and eu).

https://webcache.googleusercontent.com/search?q=cache:s7pKN1...

I want to improve my internet "speed" by not taking fucking forever to establish tcp connections to some part of it. 2m30 to push a single commit to one git host but not another. Similar times to display a static webpage on some domains.

Was the post deleted because it tells people that paying more for internet bandwidth doesn't improve latency?

Totally agree that some websites need to chill out with their javascript frameworks and start learning how to actually deliver websites rendered in the backend that reduce the amount of data people have to download.

I'm not sure I agree with that title... I just went from 15 (but it was more like 10) Mbps up, to 50 Mbps up.

Twitch tops out at 6 if you're not partnered, so having extra wiggle room means I could potentially stream online games and not run out of bandwidth.

Looks like this post was withdrawn?

... which makes it even more interesting. Not only what they wrote, but what they regretted to be published.

I just got fiber. 5-20 ms to most websites as far as I can tell. Blazing. It is the lowest bandwidth package, but the latency is still far superior to cable. And because of issues with the ISP's DNS, I am happy with 1.1.1.1 (Cloudflare).

IMO, they are correct to focus on latency instead of bandwidth as the primary factor governing perceived "speed" by the users. And speed.Cloudflare.com is one interesting tool to measure that. But it doesn't tell you anything about your latency to various other sites around the 'net.

I wonder, do folks here frequently measure that? Do you set up tools to constantly smokeping the top 100 web sites around the world, or maybe the top twenty websites that you actually use, or something else? Do you set up your own downdetector style graphs to show what latency is like to all these various services for you?

If you do set up your own kind of monitor like this, what tools do you use to do so? Is it rrdtool or Nagios or something else?

I'd be curious to know what you folks think of https://testmy.net/ and how that compares to other Speedtest type sites?

I like their comparison to other tests in other areas and other carriers, as well as their options for configuring the tests.

Anything else? Are there good toolkits out there for testing this sort of stuff?

> If the Internet is an information superhighway

Oh my.

Anyways, TL;DR:

unimportant (but somewhat understandable to Average Joe) background on how Internet works

Nonsense example witha female named Christabel

Two sentences with a corporate bullshit name to advertise what CF made a speed test site: https://speed.cloudflare.com/

Speetest looks.nice BTW and try to measure the jitter which is a thing most speed test doesn't bother.

(Link no longer works.)

unrelated but the hn web archive bots are getting redic

Seems related if the original link is dead, no?

mah, I do not see nothing special

but https://speed.cloudflare.com/ is cool

That _is_ cool

Seeing ~7% packet loss from my cellphone; is that normal?

Reply on news.ycombinator.com