Observability Stack
Metrics, logs, and traces — the three pillars of production visibility
Observability is not just monitoring — it is the ability to ask new questions about your system without deploying new code. This stack covers the three pillars (metrics, logs, traces) with both open-source and commercial options depending on your team size and budget.
Prometheus
The CNCF standard for time-series metrics. Pull-based model works natively with Kubernetes. PromQL is the lingua franca of SRE alerting.
Elastic Observability
Centralised log aggregation, distributed tracing (APM), and infrastructure metrics in one platform. Open-source core with hosted option.
Datadog
Best-in-class commercial observability. Ideal for teams that want everything pre-integrated with minimal setup. Premium pricing at scale.
New Relic
Strong APM and browser monitoring. Generous free tier makes it accessible for smaller teams before committing to enterprise contracts.
Arize AI
Monitors ML model performance, data drift, and embedding quality in production. Purpose-built for LLM and ML workloads.