DocumentaciónHelp

Cerebro

Cerebro learns from the history of your components and estimates their risk of failure over the next 1, 6 and 24 hours.

On this page

Cerebro is the ARGUNIS failure prediction module. It learns from the history of each component you turn it on for and estimates, every 15 minutes, the probability that it will fail in the next 1, 6 and 24 hours. When the risk goes up, it opens a predictive alert (“could fail”) before the failure happens.

The Cerebro menu appears if your plan includes it.

What you get

  • Failure risk per component, with the level High risk, Watch or Normal.

  • Signals that accompany the risk (which metrics are rising or falling compared to usual), so you know where to look.

  • Disk forecast: when a disk could fill up.

  • Predictive alerts that don't open incidents, so you can act before the failure.

  • A model that improves on its own with real failures and with your answers to the alerts.

  • Precursor scenarios: the state before each outage; if it repeats, Cerebro alerts you ahead of time with the cause from the previous time.

What it learns

Cerebro looks at the history of each component: CPU, memory and disk usage, availability, latency, log and HTTP errors, traffic, notifications, previous incidents, deployments, the time and day, and the state of the component it depends on. It works on servers, web and TCP services, cloud services and APM services.

Signals accompany the risk, but they are not a diagnosis of the cause. Use them as a lead to investigate.

Pages in this section

Search Argunis Monitoring

Type at least 2 characters to search Argunis Monitoring.

↑↓ to navigate · Enter to open · Esc to close