Prometheus 3.0 Unveiled: New UI, Remote‑Write 2.0, and Native Histograms

Overview

Prometheus 3.0 marks the first major release after seven years, positioning the project as a core component of the cloud‑native monitoring stack while preserving compatibility with earlier versions.

1. New UI

The default UI has been completely rebuilt, offering a modern look, reduced visual clutter, and new features such as a PromLens‑style tree view. Users can switch back to the old UI with the “old‑ui” flag. The UI now supports UTF‑8 metric and label names, though it remains in beta and may still contain issues.

2. Remote‑Write 2.0

Remote‑Write 2.0 upgrades the protocol, adding native support for metadata, samples, creation timestamps, and native histograms. String interning reduces payload size and CPU usage, and improved error handling provides more detailed client feedback.

3. UTF‑8 Support

All valid UTF‑8 characters are now allowed in metric and label names; label values already supported UTF‑8 in 2.x. When a component cannot handle UTF‑8, names are escaped with underscores. The Go client library has been updated, with other language clients to follow.

4. OTLP Integration

Prometheus 3.0 can receive OpenTelemetry metrics via a native OTLP receiver at /api/v1/otlp/v1/metrics, improving interoperability with the OpenTelemetry ecosystem.

5. UTF‑8 Normalization

With built‑in UTF‑8 support, users no longer need to rename metrics for OpenTelemetry ingestion, reducing confusion between SDK definitions and stored data. Experimental OTLP conversion options are available in the otlp configuration section.

6. Native Histograms

Native histograms provide a more efficient, lower‑cost alternative to classic histograms, using exponential bucket boundaries without requiring pre‑defined bucket selection. The feature is experimental and disabled by default; enable it with --enable-feature=native-histograms. Some aspects such as text format and accessor functions are still under development.

7. Major Breaking Changes

The release includes several breaking changes across feature flags, configuration files, PromQL syntax, and the scrape protocol. Users should consult the migration guide to understand impact and required adjustments.

8. Performance Improvements

Benchmarks on an 8‑CPU, 49 GB RAM node show reduced CPU and memory usage compared with Prometheus 2.0.0 and 2.18.0, demonstrating efficiency gains in the TSDB engine.

2.0.0 (released 7 years ago)

2.18.0 (released 4 years ago)

3.0.0 (current version)

9. Upgrade Guidance

When upgrading from 2.x, follow the migration guide and first upgrade to v2.55 before moving to v3.0. Rolling back to versions earlier than v2.55 is not supported.