It also helps avoid distributed denial of service attacks.

Internet Protocol was designed around the idea that anyone should be able to send traffic to anyone, and that that party doesn't really have control over traffic reaching them until it gets to them, where they can just drop it if they aren't interested in it. That has some benefits.

However, this meant that if someone could round up enough computers with attached bandwidth, they could send traffic to one computer, and flood its network connection. This degrades its ability to talk to the rest of the Internet.

Some people decided to start going out and—often by compromising computers in one way or another—adding them to a "botnet", which they could then use to flood servers on the Internet, making various services, mostly websites, slow or inaccessible. This is known as a distributed denial of service attack (DDoS). Some groups decided to try demanding ransoms from various websites.

If you try to flood a host behind CloudFlare, though, all you flood is CloudFlare, which is very large and has a lot of bandwidth and is going to be very difficult to bring down. You can try to reach the host behind CloudFlare via various methods—in fact, I posted something just the other day that specifically dealt with reaching that host, for a latency measurement—but if you're trying to flood them that way, then CloudFlare is in a good position to detect and block your traffic at their computers. And, I'd expect that CloudFlare will also have a lot of expertise to try to counter very, very large-scale flooding before it even reaches their systems by communicating with major telcos.

I think that what we probably need is potentially for some kind of revision to IP that permits for hosts to have a way to blacklist or whitelist traffic and push those upstream, unless we want to have something like most Internet traffic being routed through CloudFlare or similar to solve the distributed denial of service problem. We...could do that, have the Internet basically be CloudFlare, Inc, but that's not really what the designers of Internet Protocol intended, and I think that there are some good reasons not to go that route.

So basically what I gather from this is CDNs are, once again, something else that would be better for me if I just did it myself. Like most things in the tech world these days, doing it yourself is the cheapest and most freeing option. Assuming you have an old but still capable PC around, that is.

Every time I click, your post disappears and I have to reopen it.

To add to this: the issue with it comes with the implementation which, without explinaiton, requires JavaScript, web assembly, etc.

And for whatever fucking reason insists on loading several megabytes of data before even deciding whether to make you complete the challenge. This is why it's so slow.

They also do reverse tunnels, which are handy if you want to expose your website, web app, API ect to the internet without exposing your whole ass network to the internet and deal with the security implications of that, or if you don't have/don't want to have full control over the upstream local network (eg you want to stand up a web app on a weird port inside a corporate network where the ICT team won't let you do port forwarding on the router firewall) including not having a static public IP address. It also handles SSL so you don't need to install your own certs.

This lets me run websites and an API from servers in my house, without needing to do any firewall stuff on my router or having a static IP issued by my ISP. I only need one tunnel to each server, and each tunnel can handle routing all of the sites on a server to their various domain names.

Holy shit so I'm not the only one. They seem to change their UI menu names more than I change my pants.

This! Every other werk they move my tunnels to a new random spot or menu. What's up with that?