awsI saw a post on Hacker News the other day that caught my eye. It was a 'Show HN' for a system someone built to keep their apps running when a server or region goes down. The comments were full of engineers sharing war stories and debating the finer points of DNS, anycast, and health checks. It's 2024, and smart people are still having to build this kind of foundational plumbing from scratch.
That should tell you something. It tells you that the default resilience story from the major cloud providers has a gap in it. A big one.
The Promise vs. The Plumbing
We're all sold the dream of multi-region, auto-scaling, self-healing infrastructure. The marketing materials make it sound like you just check a few boxes and your application becomes invincible. But as that Hacker News thread shows, the reality is a lot more complicated.
The cloud providers give you the components—load balancers, multiple availability zones, DNS services. But they don't give you a finished product. Stitching those components together into a system that can automatically and reliably fail over during a real-world outage is your responsibility. It's a significant, ongoing engineering effort.
Why? Because the incentives aren't aligned. A provider's job is to sell you more infrastructure and services. Your job is to ensure your specific application stays online. These are not the same thing.
Your SLA Isn't an Uptime Guarantee
This brings us to the core of the mismatch: the Service Level Agreement (SLA). Most people think of an SLA as a promise of uptime. It's not. An SLA is a financial instrument that defines the penalty the provider pays if they fail to meet a certain performance level. And the penalty is usually just a small service credit.
It's a risk management tool for the provider, not a resilience guarantee for you. According to the Uptime Institute's 2022 report, more than 60% of outages cost businesses over $100,000. Your 10% service credit for the month isn't going to cover that.
The real problem is that most organizations have no systematic way to even track SLA performance, let alone enforce the contract. They rely on the provider's status page—which often lags behind reality—and a manual, time-consuming process to file for credits they're owed. Most of the time, it just doesn't get done.
What You Should Actually Be Doing Right Now
Seeing engineers build their own failover systems is a signal that the standard approach is broken. Instead of just focusing on the technical architecture, you need to fix the operational and financial accountability loop.
Here are the practical steps:
-
Stop Relying on Status Pages. You need independent, third-party monitoring that measures performance from your perspective. The only way to know if an SLA has been breached is to measure it yourself. You can't hold a partner accountable using their own data.
-
Quantify the Cost of Downtime. For each critical workload, you should have a clear number: what does one hour of downtime cost the business? This gives you the context to decide how much to invest in resilience and makes the value of SLA credits clear.
-
Automate SLA Enforcement. Manually tracking outages and filing for credits is a losing battle. It's tedious, error-prone, and gets pushed aside by more urgent work. You need an automated system that continuously monitors performance against your contract, documents any breaches, and manages the credit recovery process. This turns your SLA from a piece of paper into an enforceable agreement.
-
Connect Performance to FinOps. When you automate credit recovery, you create a direct, visible link between service degradation and its financial impact. Your FinOps team can now see the true cost of unreliability, and your engineering teams have a clear mandate to build more resilient systems.
Closing the Accountability Gap
This is the gap we built Next Signal to fill. I spent years inside large organizations watching teams struggle with this exact problem. We had great observability tools, but we were completely blind to the financial and contractual performance of our cloud providers.
Next Signal provides that missing operational layer. We independently monitor your cloud services against your specific SLAs. When we detect a breach—often before it ever hits a public status page—we automatically document it and manage the entire credit recovery process for you.
It’s not about replacing your monitoring tools. It’s about adding the financial and contractual accountability that has been missing from the cloud ecosystem. It ensures you get the performance you pay for, and when you don't, you automatically recover the cost.
Seeing smart engineers build custom failover tools is inspiring. But it's also a sign of a deeper, structural problem. Before you spin up another complex internal project, make sure you have the operational basics in place. You can't enforce a contract you aren't independently monitoring.