dpm-finder
0dpm-finder is a tool for Grafana Professional Services that analyzes Prometheus metrics to identify which ones drive high Data Points per Minute (DPM). It provides a per-label breakdown to assist in optimizing Grafana Cloud costs.
1800Updated 1mo ago
Monitoring
Quick InstallView Source
$ npx skills add https://github.com/grafana/skills --skill dpm-finderOverview
About This Skill
dpm-finder is a Python script that retrieves all metrics from a Prometheus instance and calculates their data points per minute (DPM) rate using PromQL queries. It automatically filters out histogram/summary components, Grafana internal metrics, and metrics with aggregation rules, then identifies metrics exceeding a configurable DPM threshold. The tool operates in two modes: one-time execution that outputs results to CSV, JSON, text, or Prometheus exposition formats, or as a Prometheus exporter server exposing live DPM metrics. It includes detailed logging, Docker and Docker Compose support, and is designed specifically for Grafana Professional Services to help teams diagnose and control high-volume metric ingestion.
How to Use
Install using the command: `$ npx skills add https://github.com/grafana/skills --skill dpm-finder`
Use Cases
Optimize costs in Grafana Cloud by analyzing DPM
Break down metric performances by labels
Identify critical metrics for operational efficiency
Tags
Setup & Installation
Quick Install
Terminal
$ npx skills add https://github.com/grafana/skills --skill dpm-finderAlternative Install (Git Clone)
git clone https://github.com/grafana-ps/dpm-finderRequirements
- Claude Code or compatible AI agent
Quick Start Guide
1
Install the Skill
Copy the install command above and run it in your terminal.
2
Open Your AI Agent
Launch Claude Code, Cursor, or your preferred AI coding agent.
3
Try It Out
Use the prompt template or examples below to test the skill.
4
Customize
Adapt the skill to your specific use case and workflow.
Usage Examples
Prompt Template
Analyze the Prometheus metrics for [COMPANY]'s [INDUSTRY] workload in Grafana Cloud to identify which metrics are driving the highest Data Points per Minute (DPM). Provide a breakdown by label names and values, and suggest optimizations to reduce costs. Use the following query as a starting point: `sum by (job, __name__) (rate(prometheus_tsdb_head_samples_appended_total[5m]))`. Focus on metrics with DPM > 1000.
Example Output
# High-DPM Metrics Analysis for Acme Corp's E-commerce Platform ## Top 5 Metrics by DPM | Metric Name | DPM | Labels (Top 3) | Suggested Action | |-------------|-----|----------------|------------------| | `http_requests_total` | 12,450 | `path=/api/checkout`, `status=200`, `method=POST` | Review checkout endpoint logging frequency | | `kafka_messages_consumed` | 8,200 | `topic=orders`, `partition=0`, `consumer_group=orders-service` | Reduce Kafka consumer logging granularity | | `container_memory_usage_bytes` | 6,800 | `namespace=production`, `pod=checkout-v2-abc123`, `container=checkout` | Optimize memory metric collection interval | | `http_request_duration_seconds_sum` | 5,300 | `path=/api/products`, `status=2xx`, `method=GET` | Consider sampling high-traffic endpoints | | `prometheus_notifications_total` | 4,100 | `rule_group=alerting-rules`, `alertname=HighLatency` | Review alerting rule efficiency | ## Cost Impact Analysis - Current monthly DPM: ~1.2B - Estimated cost reduction potential: 35-40% by implementing suggested optimizations - Top 20 metrics account for 85% of total DPM costs ## Recommended Next Steps 1. **Immediate**: Implement label filtering for `http_requests_total` to exclude `/health` and `/metrics` endpoints 2. **Short-term**: Reduce collection frequency for memory metrics from 15s to 30s during non-peak hours 3. **Long-term**: Evaluate metric cardinality for `kafka_messages_consumed` to identify redundant labels Would you like me to generate specific Prometheus relabeling configurations for any of these metrics?
Apply to these tools
Browse all toolsCompatible MCP servers
Browse all MCP serversFind the right skills for your stack
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan