Building an Anycast network for fun (2022)(tylerobrien.dev)
tylerobrien.dev
Building an Anycast network for fun (2022)
https://tylerobrien.dev/posts/2022/06/28/building-an-anycast-cdn
3 comments
Oh cool to hear! I didn't knew people were trying to build this - I think that's pretty fascinating, although also a difficult to pull off as a community - how do you manage malicious actors both on the ingress but also at the service layer if there is no commercial "contract". Currently those questions are even hard to address on a single-IP vHost level, so I'd be curious to hear what your thoughts are on that topic.
With the non-profit doesn't mean non-commercial, so what is your idea in that regard?
With the non-profit doesn't mean non-commercial, so what is your idea in that regard?
> With the non-profit doesn't mean non-commercial, so what is your idea in that regard?
Well in this context, I would love it if it was possible to have it non-commercial itself but perhaps what I am thinking right now is more similar to having a free tier for users to try out some free tunnels and they can realize the mission of 'being fair'. I am gonna be honest about how much it costs me and pretty much gonna just charge enough to offset that for a few months/years imo. I really don't want to earn as much or am motivated by it to be quite frank.
If the project really takes off then maybe just enough to maybe work on it full time or having a team to have enough to feed family :)
> how do you manage malicious actors both on the ingress but also at the service layer if there is no commercial "contract".
So for this, Consider the fact that I am just starting out, I suppose the answer is that I am looking out for more reasonable cloud provider who can offer good ingress support & having and talking to providers is an experience (which I enjoy!) and having people message.
Also regarding right now I think the answer probably revolves around having 1 host with cheap ingress & then scale out as such and once the project gets to a decent scale, I would assume it would have me credited enough to take on large grants from other non providers (like Nlnet) too to expand further and then have to worry about anycast itself as well
Currently, the idea is to work with low end providers themselves. They are usually more reasonable about all such stuff regarding abuse starting out given that I could be sleeping and someone can bombard. So I love talking to the company owner themselves in this context.
https://lowendtalk.com/discussion/213907/vps-with-high-bandw...
I do feel like its a bit of pitch that I am giving right now so let me try to be more honest why I felt like this.
I realized that on my netcup servers, I was over utilizing on cloudflare tunnels. I loved the idea of hooking it up and had made custom scripts which can literally just be ./expose.sh <PORT> subdomain.domain.com and it would use cf tunnels internally. Ended up deploying a lot of services with this + podman/docker.
Until recently it broke but that's aside from the matter but also I just felt like even with netcup and having an ipv4 I was using the idea of cf tunnels because I could be using it even if there was ipv6 & actually to use cf tunnels you need to migrate the domain to cf which means that you also have to have them handle domain and dns too with their nameservers.
To me combined with the fact that ~12% of internet routes through Cf & the fact that technically all of this is uncensored. It's just that we are trusting cf when we use tunnels.
Now the chances of cf abusing it is quite low but another context with which I wanted to create was that I have personally been also fascinated with trying to have alternatives to American companies.
And the whole reason I thought all of this was that I wanted to migrate my dns from cloudflare to desec.io which is a non profit german dns provider. But that would mean that I would lose all of my cloudflare tunnels. This just didn't sit right with me and made me realize of all of this and out of curiosity started wondering how much would it cost to make a cf tunnel alternative and thought of asking it on lowendtalk and figuring things out as I started wondering if an non profit idea might make sense.
To be honest, I think I should make a website about the idea :) but the issue is that I feel pretty hidden (I don't use anything other than hackernews maybe some bluesky) and like I don't have any followers or anything to go point out.
I do feel like with my frugality, I will try to absolutely minimize any costs if someone donates to me and I just like doing these things and really just want to build it because i feel like I can build it.
Do you have any suggestions for me? I would love to know from a real human being because the feeling of seeing all this hyped product and then seeing that no matter what you talk about, chances of you being not visible online are quite high is something that I am start to believe in.
Do you think that i should try to start a kickstarter perhaps? Or how do I seek donations first, I genuinely don't know and I would seriously love your feedback.
Thanks for asking me about this project and wishing you and your family a nice day miduii!
Well in this context, I would love it if it was possible to have it non-commercial itself but perhaps what I am thinking right now is more similar to having a free tier for users to try out some free tunnels and they can realize the mission of 'being fair'. I am gonna be honest about how much it costs me and pretty much gonna just charge enough to offset that for a few months/years imo. I really don't want to earn as much or am motivated by it to be quite frank.
If the project really takes off then maybe just enough to maybe work on it full time or having a team to have enough to feed family :)
> how do you manage malicious actors both on the ingress but also at the service layer if there is no commercial "contract".
So for this, Consider the fact that I am just starting out, I suppose the answer is that I am looking out for more reasonable cloud provider who can offer good ingress support & having and talking to providers is an experience (which I enjoy!) and having people message.
Also regarding right now I think the answer probably revolves around having 1 host with cheap ingress & then scale out as such and once the project gets to a decent scale, I would assume it would have me credited enough to take on large grants from other non providers (like Nlnet) too to expand further and then have to worry about anycast itself as well
Currently, the idea is to work with low end providers themselves. They are usually more reasonable about all such stuff regarding abuse starting out given that I could be sleeping and someone can bombard. So I love talking to the company owner themselves in this context.
https://lowendtalk.com/discussion/213907/vps-with-high-bandw...
I do feel like its a bit of pitch that I am giving right now so let me try to be more honest why I felt like this.
I realized that on my netcup servers, I was over utilizing on cloudflare tunnels. I loved the idea of hooking it up and had made custom scripts which can literally just be ./expose.sh <PORT> subdomain.domain.com and it would use cf tunnels internally. Ended up deploying a lot of services with this + podman/docker.
Until recently it broke but that's aside from the matter but also I just felt like even with netcup and having an ipv4 I was using the idea of cf tunnels because I could be using it even if there was ipv6 & actually to use cf tunnels you need to migrate the domain to cf which means that you also have to have them handle domain and dns too with their nameservers.
To me combined with the fact that ~12% of internet routes through Cf & the fact that technically all of this is uncensored. It's just that we are trusting cf when we use tunnels.
Now the chances of cf abusing it is quite low but another context with which I wanted to create was that I have personally been also fascinated with trying to have alternatives to American companies.
And the whole reason I thought all of this was that I wanted to migrate my dns from cloudflare to desec.io which is a non profit german dns provider. But that would mean that I would lose all of my cloudflare tunnels. This just didn't sit right with me and made me realize of all of this and out of curiosity started wondering how much would it cost to make a cf tunnel alternative and thought of asking it on lowendtalk and figuring things out as I started wondering if an non profit idea might make sense.
To be honest, I think I should make a website about the idea :) but the issue is that I feel pretty hidden (I don't use anything other than hackernews maybe some bluesky) and like I don't have any followers or anything to go point out.
I do feel like with my frugality, I will try to absolutely minimize any costs if someone donates to me and I just like doing these things and really just want to build it because i feel like I can build it.
Do you have any suggestions for me? I would love to know from a real human being because the feeling of seeing all this hyped product and then seeing that no matter what you talk about, chances of you being not visible online are quite high is something that I am start to believe in.
Do you think that i should try to start a kickstarter perhaps? Or how do I seek donations first, I genuinely don't know and I would seriously love your feedback.
Thanks for asking me about this project and wishing you and your family a nice day miduii!
Lowendtalk is a great community to talk about in such things and buyvm (frantech) or francessco is quite active in the community too.
I have even talked to francesco in their discord themselves iirc. It was fun and I didn't understand the hype behind BuyVM but the community or the discord server, people love francesco a lot.
Great read. To be honest regarding my cf tunnel idea. I feel like its more of an economies at scale/loss making idea & only cf is able to profitably execute it but my intention was that cf runs ~12% of the data around the world and I just wished if a non profit could've been established regarding a cf tunnel idea and I even talked to providers about it.
But I felt like I needed funding and so I did drop the idea unfortunately. I wanted to fund the project myself but I felt like I was incredibly frugal and the idea of funding/sustaining it myself was quite infeasible.
just out of Curiousity, curious to hear what the hn communities think about an non profit cf tunnel.