dt-obs-services

This skill provides guidance for monitoring application-service performance and health in Dynatrace using DQL (Dynatrace Query Language). It centers on RED metrics (Rate, Errors, Duration) and runtime-specific telemetry, helping engineers answer questions about service response times, error rates, throughput, SLA compliance, messaging health, and service-mesh overhead without hand-writing queries from scratch.

The skill documents the key service metrics (request response_time, count, and failure_count in microseconds/counts) and supplies ready-to-adapt DQL query templates for response-time percentiles (p50/p95/p99), error-rate calculation, failure spike detection, failures by HTTP status, throughput and peak-traffic detection, traffic growth, performance-degradation detection against a baseline, and Kubernetes-context breakdowns by workload, namespace, and cluster. Additional sections cover span-based advanced analysis for custom SLA thresholds and health scoring, messaging metrics for publish/receive/process rates and consumer-lag detection, and service-mesh ingress monitoring including mesh-vs-direct overhead. The skill is explicitly scoped: it defers infrastructure metrics, logs, and tracing to sibling dt-obs skills.

Target users are SRE and platform teams, application-performance engineers, and on-call responders who operate services instrumented by Dynatrace, particularly across Java, .NET, Node.js, Python, PHP, and Go runtimes and Kubernetes environments. Typical use cases include diagnosing latency regressions, tracking SLA compliance, spotting error or failure spikes, comparing multi-cluster performance, and analyzing queue/topic messaging health. All operations are read-only observability queries.