Application performance signals

New Relic measures response times, throughput, and error rates across your services. An agent reads these signals against the thresholds your team approved and raises an alert when a service degrades, naming the affected transaction and the metric that moved. It records the reading in the incident note so responders see the trend, not just a snapshot. Normal fluctuations inside the agreed band pass without noise. When a signal is ambiguous, a brief spike that may resolve itself, the agent holds and re-reads before escalating, and unclear patterns are handed to an engineer to interpret.

Alert incidents

New Relic opens incidents when a condition your team defined is breached. An agent reads a new incident, applies the routing rule you approved, and notifies the owning team with the policy, severity, and first readings attached. It updates the incident record as it groups related alerts so responders are not paged twice for one root cause. Incidents that match no known owner, or that combine several services in a way the rule does not cover, are escalated to a human coordinator who decides ownership rather than letting the alert sit unassigned.

Infrastructure metrics

New Relic reports host and container metrics like CPU, memory, and disk. An agent reads these against the limits your team set and flags a host trending toward exhaustion, updating a ticket with the resource, the current value, and the recent slope. Routine load that stays within range is left alone. When a metric crosses a limit but the service still responds normally, the agent notes the mismatch and asks an engineer whether to act, since scaling or restarting infrastructure is a decision people should own rather than an automatic reaction to one reading.