Grafana Labs introduces the Graphviz panel, currently in private preview for all Grafana editions, which lets users describe workflow and network diagrams using the DOT graph-description language instead of manually placing boxes and arrows. The panel binds live Grafana data to diagram nodes and edges, enabling use cases like payment pipelines, service maps, network weathermaps, on-call runbook decision trees, and CI/CD pipeline visualizations that recolor and relabel themselves as underlying metrics change. It supports builder mode, code mode, and a data-driven query mode where DOT strings are generated from query results, plus integration with Grafana Assistant to generate DOT from natural language. Installing the plugin requires Grafana 12.3 or later.
Table of contents
The gap we wanted to addressGraphviz: from manual drawing to describing the diagramQuestions this post answers
What is the Graphviz panel in Grafana and what does it require to install?
The Graphviz panel is a new Grafana panel type, currently in private preview across all editions, that renders diagrams described in the DOT graph language and binds them to live Grafana data. It replaces manual diagram placement (as required in the canvas panel) or the unmaintained Angular-based FlowCharting plugin. Installing it requires Grafana version 12.3 or later, available through the Grafana plugin catalog. Developers tracking new Grafana panel releases can follow rollout details on daily.dev.
What DOT layout engines does the Grafana Graphviz panel support?
Four layout engines are supported: dot for hierarchical top-down or left-right flows, neato and fdp for organically arranged interconnected networks, and circo for circular or radial relationships. Each suits different diagram types, for example dot for payment pipelines and runbooks, and neato for network weathermaps where geographic layout matters. Teams choosing a diagramming approach for dashboards can compare layout options like this on daily.dev.
How can I bind live metrics to a diagram in Grafana instead of hardcoding values?
The Graphviz panel offers several data-binding mechanisms: color by threshold to map a node's fill to a metric like CPU or success rate, label templates using ${field} syntax for live values, edge width scaled to throughput via penwidth, rich tooltips and data links for drill-down, and dashboard variables like ${service} to parameterize an entire diagram across teams. Engineers building live, data-driven dashboards can dig into approaches like this on daily.dev.