Ask HN: Reasonable ways to avoid single points of failure in web services?
I previously commented [1] that I couldn't log into any of the 5 monitoring services I have setup, due to the current Cloudflare outage.
Now in principle, even with this kind of outage, it should be possible to log into your domain registrar, reset your nameservers and add some DNS records to get atleast some parts of your infra back online by bypassing CF.
Guess what? Not only can't I log into Cloudflare, but also Porkbun, Dynadot and Spaceship (owned by Namecheap).
It seems to me that the internet is getting far too centralized. A single outage should not mean everything is down, we saw this with AWS recently and now CF.
Is there a reasonable way to get the benefits that Cloudflare provides like DDoS protection, CDN, turnstile, etc. without using CF (or an equivalent service) and making it a single point of failure?
[1]: https://news.ycombinator.com/item?id=45964199 That's why Blockchain has been invented.