The Illusion of Convenience and Guarantees
There was a moment in time where serverless computing seemed to be everywhere (it’s still around, but not much of a buzzword anymore). It promises the ability to run applications without having to worry about infrastructure management. No servers to maintain, no scaling nightmares, no hardware failures to keep you up at night. Just deploy your code, and the cloud handles the rest.
Sounds familiar? It should. It’s the same pitch car-sharing services like Cambio or Poppy make: “Why own a car when you can just use one when you need it?” No maintenance, no insurance headaches, no need to worry about depreciation. Just book and drive.
Let’s call it carless, the transportation equivalent of serverless. But much like serverless computing, carless transportation comes with trade-offs—ones we don’t always think about until it’s too late.
The SLA Paradox: Owning vs. Leasing
When we own infrastructure—whether it’s servers or a car—we demand high availability. If we run a critical business application, we expect our data center providers to guarantee 99.99% uptime because anything less could mean lost revenue, angry customers, and a bruised reputation.
Yet, when we move to the cloud, we accept something different. Not a single public cloud provider offers 99.99% uptime on their infrastructure. Their SLAs (Service Level Agreements) are lower because, ironically, their scale works against them. There are simply too many moving parts, too many regions, and too much complexity for them to commit to the same availability we demand from our on-prem partners.
And yet, we blindly assume we’re getting better availability because the cloud feels “big” and “automated.”
What Happens When the System Fails?
Let’s switch back to our car analogy.
If I own a car, I expect it to be ready whenever I need it. If something breaks down, I’ll do whatever it takes to get it fixed immediately—because my mobility depends on it.
Now, let’s say I rent a car. In theory, I should have the same reliability, right? After all, I’m paying for a service. But here’s where things get interesting.
Imagine you’re on a road trip to the south of France, and your company-leased car breaks down. You call the leasing company, expecting an immediate replacement, only to hear:
“We’ll get to it when we get to it.”
Or worse:
“We’re sorry, but our roadside assistance partner is currently at capacity. We appreciate your patience.”
Suddenly, your 99.99% mobility expectation is shattered. The leasing provider, like a cloud provider, has different priorities than you do. They optimize for scale, cost efficiency, and service contracts—not for your personal urgency.
You, my friend, are stuck at a gas station somewhere between Marseille and Montpellier, wondering why you didn’t just bring your own car with a more expensive (but ultimately better) maintenance contract.
The Fine Print No One Reads
One of the biggest misconceptions about cloud SLAs (and something I often have to explain to customers) is that they don’t actually guarantee uptime—they guarantee compensation. If a cloud provider goes down, you might get service credits, but that doesn’t help when your app is offline and your customers are furious.
Similarly, a leasing company might cover repairs or even provide a temporary replacement eventually, but if your trip is disrupted, you’re out of luck.
Would you accept a car that only works 99% of the time? Probably not. But we accept cloud providers with lower guarantees because we assume they know best.
Choose Wisely
Neither serverless nor carless is inherently bad—they both offer convenience, flexibility, and the ability to offload operational headaches. But don’t let the marketing fool you.
- If you truly need 99.99% uptime, be prepared to architect for resilience, because your cloud provider won’t guarantee it.
- If you truly need a car available at all times, consider owning one, because leasing companies have their own priorities.
At the end of the day, you can rent the infrastructure, but you can’t rent responsibility.
Choose wisely.