By Pauline Shangett, Chief Technique Advisor at NOWNodes
On October 20, 2025, the web faltered. For hours, numerous apps, platforms, and providers merely stopped working. Fortnite froze. Snapchat crashed. Alexa went silent. Even main banking and buying and selling apps had been down.
The trigger wasn’t a cyberattack or a hack; it was a routine software program replace gone unsuitable in Amazon Internet Companies’ US-EAST-1 area, probably the most relied-upon items of digital infrastructure on the planet.
A small configuration change led to DNS failures that rippled throughout the worldwide internet, breaking every thing from gaming to monetary providers to components of the crypto ecosystem.
It was a second of silence and a reminder that the “cloud” isn’t some ethereal, distributed community. It’s a group of knowledge facilities owned by a couple of corporations. And when one in every of them sneezes, the web catches a chilly.
For a lot of the world, the outage was an inconvenience. For Web3, it was an existential warning.
Centralized comfort, decentralized illusions
The trendy web runs on comfort. Platforms like AWS, Google Cloud, and Azure have made it a lot simpler for corporations to scale. Startups don’t want to purchase racks of servers or run their very own information facilities anymore. You simply pay for what you employ, get an occasion working in a couple of minutes, and may give attention to constructing your product fairly than coping with {hardware}.
However this comfort comes with a catch: reliance. When your infrastructure lives someplace else, you’re placing numerous belief and management into another person’s arms. The Web3 area, the world’s loudest advocate for decentralization, nonetheless leans closely on centralized infrastructure. Many DApps, RPC endpoints, wallets, and validator nodes run on the identical few suppliers, usually in the identical areas.
If a single cloud area fails, whole “decentralized” ecosystems grind to a halt. The irony is painful: we’re decentralizing finance and governance however centralizing the servers that preserve them alive. When AWS goes down, it’s not only a matter of downtime; it additionally damages belief. If a decentralized system can’t stand up to a single level of failure, can it actually be referred to as decentralized?
The hidden price of centralization
Centralized infrastructure concentrates not solely threat but in addition management. Cloud suppliers can, and do, throttle, droop, or reprice providers at will. They function as invisible intermediaries with the ability to have an effect on every thing from latency to liquidity.
For years, cloud computing was cheaper and extra versatile than proudly owning {hardware}. However because the “Massive Three” clouds consolidated dominance, the market started to look much less like innovation and extra like oligopoly.
In 2024-2025, AWS compute prices elevated by over 20%, with almost 40% of corporations reporting invoice spikes exceeding 25%. The identical providers that when enabled startup agility now punish success with unpredictable scaling charges.
And when your product’s uptime and monetary runway depend upon a single supplier’s enterprise mannequin, you’re not in management. You’re a tenant.
{Hardware} returns, not as nostalgia, however necessity
Proudly owning servers may sound outdated, however in 2025, it’s turning into a strategic benefit.
The mathematics isn’t too sophisticated. A bodily server costing about $1,100, unfold over ten years, comes out to roughly $110 a month. Examine that to cloud computing at scale, which might simply hit $2,000 to $7,000 a month. However the actual profit isn’t nearly cash. While you run your personal {hardware}… effectively, you’re actually the one in cost. You get to determine the place the information lives. You determine how redundancy ought to work. You possibly can tweak issues for velocity, for safety, or for compliance. No ready for a cloud supplier to roll out some new characteristic. No coping with their API limits. You simply do it your approach.
And importantly, your service doesn’t vanish as a result of one cloud area had a nasty morning. Proudly owning your infrastructure doesn’t imply going absolutely offline or constructing bunkers filled with machines. It means designing for distribution, spreading your techniques throughout suppliers, geographies, and {hardware} fashions so no single failure can take every thing down.
So it’s not “cloud vs. metallic.” It’s about management vs. fragility. Clouds will fail. {Hardware} can fail. However when techniques are distributed, redundant, and supported by actual engineers who perceive failure as an inevitability, the general structure turns into antifragile.
Designing for failure: the distributed mannequin
At NOWNodes, we design infrastructure round one assumption: failure will occur. That’s why our structure is globally distributed throughout the European Union, the US, and Asia, with information facilities in Germany, Finland, the Netherlands, the U.S., and Singapore. Every location is chosen not only for community efficiency however for political stability and operational security.
Each crucial system follows a 2N+1 redundancy mannequin. Which means if one system fails, one other takes over immediately. If two fail, site visitors nonetheless routes by way of the third. Downtime isn’t “prevented”; it’s absorbed.
We additionally take a look at failure on function. Our engineers run managed outages in mirrored environments to establish weak factors earlier than they break in manufacturing. Resilience doesn’t come from assumptions, it comes from really working towards it. And it’s not nearly machines; infrastructure entails individuals, processes, and numerous little issues working collectively.
When the AWS outage hit, 1000’s of groups scrambled in panic. Who might they name? What might they do? In Web3, even a brief downtime can freeze transactions or lock individuals’s funds. That’s why we don’t simply depend on chatbots or lengthy ticket queues. Our engineers are on name on a regular basis on Slack, Telegram, and dwell chat and often reply in only a few minutes.
Most points are resolved inside hours, not days. As a result of resilience isn’t nearly techniques staying up; it’s about individuals staying current once they go down.
Infrastructure is belief
Infrastructure isn’t seen till it fails, proper? Customers don’t take into consideration the place their transactions are processed or how their pockets connects to a blockchain till it stops working. And when that occurs, belief erodes immediately.
The AWS outage was a reminder of that invisible belief relationship between platforms and customers. Even “good” units couldn’t escape it. There was a viral publish a few good mattress that stopped working as a result of AWS went down. Sounds ridiculous, but it surely’s really an ideal metaphor. The extra linked and “good” our world will get, the larger the mess when centralized techniques fail.
That’s why decentralizing infrastructure isn’t nearly ideology; it’s about performance. It’s about guaranteeing your product, your blockchain, or your “good mattress” retains working when the web’s greatest supplier takes a nap.So sure, use NOWNodes, and your good mattress gained’t crash when AWS does.