With the memories of the latest AWS outage still fresh in our heads, we were treated to another web outage today.

While credits for disrupting a significant portion of the internet usually go to Amazon's infamous "US-East-1" region, Cloudflare reminded us not to forget about them either.

When a few services rule the net

The vast market share of services like Amazon Web Services, Cloudflare, Microsoft Azure, and others becomes abundantly clear when one of them goes down.

Not only do websites go down, but other internet-reliant services like chat applications, collaboration tools, music streaming platforms, AI agents, and more do as well.

Damage varies from complete service disruptions to valid excuses for folks to touch grass again.

A legacy of "Here we go again"

This is not new behavior, either.

I cannot remember the first time I observed "US-East-1" day, but it surely was when Twitter was still Twitter and no one was considering otherwise.

(And yes, that is a long time ago now.)

Every year these major outages happen, and every year we make jokes or complain, but do we learn from them? Probably not.

I mean, I do know that there are ways to fall back on non-broken services when one goes down, but I couldn't exactly tell you about all those strategies either.

Security concerns for Cloudflare sites?

I saw an interesting take on the Fediverse earlier today (which I sadly failed to bookmark), where someone wondered:

What about those sites that use Cloudflare to keep unwanted traffic out, and never built security into their own code?

The follow-up argument was that some companies disabled Cloudflare to reopen their platforms. If those companies did not take sufficient security measures, they would become easy targets for malicious parties.

(And if not malicious parties, then at least some "oops" moments from developers or creative customers.)

The broken services are not all to blame

Whether it is AWS, Cloudflare, Azure, or another service, sometimes something breaks and takes down half the internet with it.

And to be clear, the problem is not solely with these services: we around the world who rely on a commercial US company for our daily bytes are also part of the problem.

Not to mention that, given that these platforms are massive, it is actually quite impressive that they do not go down more often.

Additionally, since they are already a staple in many products, there is less financial incentive for companies to provide high-availability services.

But we have been going over these issues for what, fourteen (14) years now? In all those years, we still haven't decentralized the internet much.

Wrap-up

I do not know where I am going with this story, but there you have it:

The internet is down, and that is not a good thing, but we seem to have accepted it as an inevitability rather than a problem to solve.

Maybe enjoy this outage by performing neglected, offline activities?

Appendix A: AWS outage overview

To my delight, I just discovered that Wikipedia keeps a list of AWS outages, which confirms that they have a solid track record of breaking the internet once a year.

Except for last year, when there was no major outage... slackers.