New Relic Reliability Engineer key terms

A detailed record of one slow or sampled request showing each segment, its duration, and where time was spent across services and database calls.

A user-satisfaction index from 0 to 1 that classifies response times as satisfied, tolerating, or frustrated against a configured target threshold T.

A timed unit inside a transaction trace, such as a database query, external call, or method, used to attribute latency to code or dependencies.

Correlating a single request across multiple services by propagating trace and span identifiers so end-to-end latency and call paths can be reconstructed.

One timed operation within a distributed trace, carrying a name, start time, duration, attributes, and parent relationship to form a call tree.

Passing standardized trace identifiers between services and queues via headers so downstream work joins the same trace; W3C traceparent is the common format.

The share of transactions or requests completed with an error, typically expressed as a percentage over a chosen time window and filtered by service.

A grouped view of exceptions or error classes across a service, showing occurrence counts, affected transaction names, and representative stack traces for triage.

A target for a measured reliability indicator, such as the percentage of requests faster than a threshold, used to track compliance over rolling windows.

Agent-based instrumentation of real user page loads in the browser, capturing core web vitals, JavaScript errors, AJAX calls, and session-level timing.

Google-defined page experience metrics including Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift, used to judge perceived load and responsiveness.

Scripted checks that run from managed locations on a schedule to test availability and response time of endpoints or full browser journeys independent of real traffic.

A reconstructed playback of a real user's browser session with DOM changes, clicks, and console events, useful for diagnosing errors and friction in context.

Configuration that hides or blocks sensitive page content, inputs, or text from browser telemetry and session replay to protect user privacy.

Instrumentation of iOS and Android apps that captures crashes, network requests, screen loads, and handled exceptions from real user devices.

Platform capability that correlates anomalies across services and infrastructure and surfaces probable root causes rather than isolated threshold breaches.

Statistical baseline modeling of a metric's normal behavior over time so alerts fire on unusual deviation rather than a fixed static threshold.

An alert condition comparing a metric against a fixed value, such as error rate above two percent, with no learned baseline behavior.

A container that groups alert conditions and defines incident creation behavior, including rollup strategy and notification workflow linkage.

A single rule evaluated against a NRQL query or metric stream that determines when a signal breaches its threshold and opens an incident.

The query language for telemetry in the platform, using SELECT, FROM, WHERE, FACET, and TIMESERIES to aggregate events, metrics, and spans.

A shareable collection of charts and query widgets arranged in pages, often filtered by variables, used to visualize service and infrastructure health.

A named grouping of entities, such as services and hosts, that aggregates their health signals into one view for ownership and alerting.

A monitored object in the entity model, such as an application, host, or database, identified by a GUID and carrying tags and golden metrics.

A small standard set of health signals per entity type, such as throughput, error rate, and response time, used for consistent cross-service comparison.

Software installed on a host that collects CPU, memory, disk, network, and process metrics and forwards them to the platform.

A packaged collector for a specific technology, such as a database or message queue, that emits standardized metrics and inventory data for that system.

Vendor-neutral specification and SDK set for traces, metrics, and logs, supported as an ingestion path alongside proprietary agents.

A configured output in an OpenTelemetry collector or SDK that ships telemetry to a backend endpoint using the OpenTelemetry Protocol.

An OpenTelemetry component with receivers, processors, and exporters that receives, batches, filters, and forwards telemetry to one or more backends.

Shipping log files or streams from hosts and containers into the platform, commonly via an agent plugin or collector, so they can be parsed and queried.

A configuration that extracts structured attributes from raw log lines using patterns such as Grok, enabling filtering, faceting, and alerting on log content.

A clustered signature that groups similar log messages by shape, helping identify the most frequent or newly appearing messages without reading each line.

The configured period telemetry remains queryable, varying by data type such as metrics, events, traces, and logs, and affecting historical analysis.

The volume of telemetry accepted by the account, typically billed by data type and used to reason about cost and sampling decisions.

Deliberately recording only a portion of traces or events to control ingest volume while preserving statistically useful latency and error signals.

Identifying, scoring, and tracking security weaknesses in libraries and container images, with prioritization based on severity and runtime exposure.

Ranking security findings by combining severity with runtime context such as whether the affected component is actually executing and internet-facing.

Centrally managing agent versions and configurations across many hosts, including staged rollout of upgrades and config changes to groups of agents.