From Natural Language to PromQL: A Catalog-Driven Framework with Dynamic Temporal Resolution for Cloud-Native Observability

2026-04-16 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Data Science & Analytics · Depth: Advanced, extended

Summary

A catalog-driven framework translates natural language questions into PromQL queries for cloud-native observability, addressing the complexity of Prometheus metrics. This system features a hybrid metrics catalog, combining a static base of approximately 2,000 metrics with runtime discovery of hardware-specific signals across NVIDIA, Intel Gaudi, and AMD ROCm GPUs. A multi-stage query pipeline performs intent classification, category-aware metric routing, and multi-dimensional semantic scoring. Crucially, it includes a dynamic temporal resolution mechanism that interprets diverse natural language time expressions into appropriate PromQL duration syntax. The framework integrates with the Model Context Protocol (MCP) for LLM interactions, achieving sub-second metric discovery and a full pipeline completion time of approximately 1.1 seconds via the catalog path. It is deployed on production Kubernetes clusters managing AI inference workloads, supporting natural language querying across cluster health, GPU utilization, and model-serving performance.

Key takeaway

For NLP Engineers building observability tools, this framework demonstrates a robust approach to natural language-to-PromQL translation. You should consider adopting a hybrid metrics catalog and a multi-stage pipeline with dynamic temporal resolution to handle the complexity of cloud-native metrics and diverse time expressions. This can significantly reduce the burden on SREs and improve query accuracy in production environments, especially for AI inference workloads.

Key insights

A catalog-driven framework translates natural language into PromQL, simplifying cloud-native observability with dynamic temporal resolution.

Principles

• Hybrid catalogs balance static knowledge with dynamic discovery.
• Multi-stage pipelines decompose complex NL-to-query tasks.
• Temporal resolution is critical for time-series query correctness.

Method

The system uses a hybrid metrics catalog, a multi-stage query pipeline for intent detection and metric selection, and a dynamic temporal resolver to map natural language time expressions to PromQL syntax, integrated via MCP.

In practice

• Use a hybrid catalog for metrics with dynamic components.
• Implement multi-stage processing for NL-to-DSL translation.
• Prioritize dynamic temporal resolution in time-series querying.

Topics

• Natural Language to PromQL
• Cloud-Native Observability
• Hybrid Metrics Catalog
• Dynamic Temporal Resolution
• Large Language Models

Best for: NLP Engineer, AI Scientist, Research Scientist, MLOps Engineer, AI Engineer, DevOps Engineer

Related on AIssential

• Amazon CloudWatch Introduces OpenTelemetry Metrics Support in Preview — CloudWatch now natively ingests OpenTelemetry metrics, completing its observability suite and simplifying metric collection.
• Building Declarative Data Pipelines with Snowflake Dynamic Tables: A Workshop Deep Dive — Declarative data pipelines with Snowflake Dynamic Tables automate ETL, reducing complexity and manual orchestration.
• Ask Your Factory Floor Anything: Structuring Industrial Data for AI Agents — Integrating time-series data and equipment context in a single PostgreSQL database simplifies contextual queries and enhances data integrity.
• Text-to-SQL solution powered by Amazon Bedrock — Text-to-SQL solutions using Amazon Bedrock democratize data access by translating natural language questions into actionable insights.
• Building an Intelligent Natural Language to SQL Pipeline Using LangGraph and Multi-Agent… — A multi-agent LangGraph architecture enhances NL-to-SQL accuracy and safety by decomposing tasks and integrating human oversight.
• A Calculus-Based Framework for Determining Vocabulary Size in End-to-End ASR — A calculus-based framework formally determines optimal vocabulary size for end-to-end ASR systems.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.