Cloudflare went down.
A month or two ago it was an AWS region.
We have centralized a large portion of our Internet Infrastructure. In a lot of ways it made sense. Amazon, Microsoft, Akamai, Cloudflare and others provided a needed service. There is a reason we use these services. Most of the time they do the job and they do it well. Depending on your price point you can more easily manage your DNS/CDN needs with Cloudflare, and host your services and apps on AWS, Azure, and Akamai without real problems.
We use them because they for the most part work. It’s easier and cheaper than trying to build, maintain, and host a solution locally. If you are like me then setting up the needed infrastructure to run a few for fun websites and passion projects isn't worth it. Its easier with tools that work.
Unfortunately, when they break you are out of luck really quickly.
With how much of the Internet is hosted and routed through a few key players, the reasonable question becomes why isn't there more competition in the space? Cloudflare alone killed multiple sites including X, Archive of Our Own, OpenAI, Uber, The Google Store, League of Legends, and many others my sites included. With so much shut down due to this particular single point of failure, one has to ask, are there alternatives? Are we really meant to be so dependent on a few companies for critical infrastructure?
The answer is, there are alternatives. However, companies like Cloudflare got so big because they offered a good service at a fair price. Even the free tier provides significant improvements on site security and ease of DNS configuration. And that has lead to a major number of people including myself being hit when one service fails. And I have thought of moving my DNS records over to Akamai to keep everything centralized. Yes, it’s a big single point of failure, but if my VPS dies then everything dies.
The real question is, is there anyway to setup anything that could bypass such an issue? Any one things hits me at any level and I am stuck until its sorted. But is it doable to have contingencies in the event any of these things bust? In a way that’s not going to toss a small operation like mine into crippling debt? No clue.
Maybe the reality is that the services we depend on are the ones who need to ensure they have the appropriate contingency plans.
It kinda sucks knowing that, at the end of the day you have to trust in those who have responsibility over critical infrastructure to not break it.
Comments