Cloud Native 9 min read

What’s New in Prometheus 3.0? UI Overhaul, Remote Write 2.0, UTF‑8 & OTLP Support

Prometheus 3.0, the first major release in seven years, introduces a revamped UI, Remote Write 2.0 with native metadata and histogram support, full UTF‑8 metric and label names, OTLP ingestion, performance gains over 2.x, and a roadmap of upcoming cloud‑native enhancements.

dbaplus Community
dbaplus Community
dbaplus Community
What’s New in Prometheus 3.0? UI Overhaul, Remote Write 2.0, UTF‑8 & OTLP Support

New User Interface

Prometheus 3.0 ships with a completely rewritten UI that is enabled by default. The UI uses a modern component stack, includes a PromLens‑style tree view, and supports UTF‑8 metric and label names. Users can temporarily switch back to the legacy UI with the old-ui feature flag.

Prometheus 3.0 UI screenshot
Prometheus 3.0 UI screenshot

Remote Write 2.0

Remote Write 2.0 extends the previous protocol with native support for:

metadata

exemplars

creation timestamps

native histograms

It adds string interning to reduce payload size and CPU usage during compression/decompression, and improves handling of partial writes by returning detailed error information.

Specification: https://prometheus.io/docs/specs/remote_write_spec_2_0/

UTF‑8 Support

Prometheus 3.0 accepts any valid UTF‑8 characters in metric names, label names, and label values, matching the behavior of the 2.x series. Exporters must emit UTF‑8 names; otherwise the legacy underscore‑escaping is applied. Queries can use quoted identifiers to reference UTF‑8 metrics, and the __name__ label can be set manually. Currently only the Go client library implements UTF‑8 support; other language clients will follow.

OTLP Support

Prometheus 3.0 includes a native OTLP receiver at the endpoint /api/v1/otlp/v1/metrics. Configuration follows the OTLP section of the Prometheus config reference.

Guide: https://prometheus.io/docs/guides/opentelemetry

Configuration reference: https://prometheus.io/docs/prometheus/latest/configuration/configuration/#:%7E:text=Settings%20related%20to%20the%20OTLP%20receiver%20feature

Native Histograms

Native histograms are a new metric type that uses exponentially increasing bucket boundaries, eliminating the need for manual bucket selection. The feature is experimental and disabled by default; enable it with: --enable-feature=native-histograms Text format support and related functions are still under design.

Breaking Changes & Migration

Version 3.0 introduces several breaking changes:

New feature flags (e.g., old-ui, native-histograms)

Configuration file syntax adjustments

PromQL updates (quoted identifiers for UTF‑8)

Scrape protocol modifications

Upgrade path: first upgrade to Prometheus 2.55, then to 3.0. Rolling back from 3.0 to 2.55 is supported; rollback to earlier 2.x releases is not.

Migration guide: https://prometheus.io/docs/prometheus/3.0/migration/

Performance Improvements

Benchmarks using the prombench macro‑benchmark suite show lower CPU and memory usage for the TSDB mode in 3.0 compared with 2.0 and 2.18 on an 8‑core, 49 GB node.

Performance chart
Performance chart

Benchmark source: https://github.com/prometheus/prometheus/pull/15366

References

Remote Write 2.0 spec: https://prometheus.io/docs/specs/remote_write_spec_2_0/

OTLP guide: https://prometheus.io/docs/guides/opentelemetry

OTLP configuration: https://prometheus.io/docs/prometheus/latest/configuration/configuration/#:%7E:text=Settings%20related%20to%20the%20OTLP%20receiver%20feature

No‑suffix implementation proposal: https://github.com/prometheus/proposals/pull/39

Migration guide: https://prometheus.io/docs/prometheus/3.0/migration/

prombench macro‑benchmark: https://github.com/prometheus/prometheus/pull/15366

Binary downloads: https://prometheus.io/download/#prometheus

Container images: https://quay.io/repository/prometheus/prometheus?tab=tags

cloud-nativeobservabilityPrometheusUTF-8remote_writeOTLP
dbaplus Community
Written by

dbaplus Community

Enterprise-level professional community for Database, BigData, and AIOps. Daily original articles, weekly online tech talks, monthly offline salons, and quarterly XCOPS&DAMS conferences—delivered by industry experts.

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.