CNCF: 46.7% of cloud-native teams still run 2–3 parallel observability stacks

How many stacks are teams actually running?

CNCF’s (Cloud Native Computing Foundation) February survey of 407 cloud-native professionals, published on May 6, 2026, on the CNCF blog, shows that 46.7% of organizations run two or three observability tools in parallel. Only 7.4% have achieved unified observability — a single platform for all telemetry signals.

Observability is the practice of monitoring the state of distributed systems by combining logs, metrics, and traces (records of how requests flow through microservices).

What are the barriers to unification?

The primary barriers are not a lack of tools but operational friction. The survey identifies three peaks:

• 54% of respondents cite dashboard and alert configuration as the top challenge.
• 46.4% have trouble with integration between tools.
• 33.2% struggle with setting up data pipelines — the flow for collecting, transforming, and storing telemetry.

OpenTelemetry, a vendor-neutral open-source standard for collecting observability data, is identified as the “strongest lever for composition and interoperable observability.” Alongside it, Prometheus leads for metrics, Jaeger and Tempo for traces, and Fluentd and Loki for logs.

What about AI adoption?

Although 59.5% of respondents want AI-powered anomaly detection — automated detection of anomalies in telemetry — as many as 48.3% require human approval before any autonomous remediation action. The finding reflects the broader industry caution toward fully autonomous operations in production.

While 81% report satisfaction with current tools, 63% are open to switching; integration quality (55.5%) is the deciding factor over individual features.