Operations | Monitoring | ITSM | DevOps | Cloud

Netdata has an incredibly powerful alerting engine. But this can sometimes be a double-edged sword: the flexibility to build incredibly specific, intelligent alerts is immense, but mastering its syntax can feel like learning a new language. We’ve heard this from so many of you. You tell us that configuring alerts is often the steepest part of the learning curve, a task that falls to the one “Netdata expert” on the team who has spent the time digging through the documentation.

Regular monitoring practices can emphasize application response time, but queue time is also often an early and important warning sign. If it rises, you’ll quickly see downstream effects: tail latency, timeouts, and error spikes. This means that this metric can give you a head start tackling app issues before they become user problems. In this post, we’ll discuss queue time, how things can go off track, and practical steps to turn it around.

By Doreen Jacobi, DERDACK / SIGNL4 It is that time of the year again. Time to reflect and look back at 2025. And I find myself thinking less about platforms and features – and more about the people behind them. The engineers who pick up the phone at 2 a.m. The operators who make judgment calls with incomplete information. The responders who keep systems running when everything feels urgent. If this year taught us anything, it’s this: technology can detect the problem, but people solve it.

Get alerted and fix issues with full context. Qovery Observe notifies you when something goes wrong and guides you straight to the metrics and signals that explain why, all in one place.

Disclosure: This comparison is written by our product marketing team that works closely with IT operations and on-call workflows. While we build incident alerting software ourselves, this guide is designed to help teams understand how different tools fit different operational needs. We believe there is no single “best” tool. Only the right fit for a given team.

In Part 1, we talked about all the hidden complexity inside AI systems: the pipelines, GPUs, embeddings, vector databases, orchestration layers, and everything else that quietly determines how reliable an AI-first product really is. But all of that software still rests on something far less glamorous: the physical infrastructure underneath it.

“We’ve got a new alert – have you seen it yet?”“Which one? The CPU spike or the unusual login?”“The login. Same region as yesterday. But the CPU thing looks suspicious too.”“…Alright, I’ll check the firewall logs. You take the containers.”“Perfect. Let’s hope this doesn’t turn into another all-hands situation.” Does this conversation sound familiar?