ER WikiWhen ER diagrams become spaghetti
Small schemas are easy to visualize. As more tables share keys and cross business domains, finding the relationship you need becomes harder than drawing it.
EDA-style routing for database relationships
Readable ER diagrams for complex database schemas.
ER Wiki applies orthogonal routing, shared buses, and obstacle avoidance to database relationships — helping growing schemas stay navigable instead of turning into spaghetti.
Follow the wire, not a wall of lines.Open the demo. Hover to trace a relationship; double-click a wire to read its meaning.
Route relationships like circuits
Circuit-design tools have dealt with dense connectivity for decades. ER Wiki borrows that perspective: organize paths, shared connections and reading scopes — not just the table positions.
EDA × Database
Why EDA-style routing?
Database schemas and circuit schematics share a visualization problem: many nodes, many connections, limited 2D space. ER Wiki focuses on how relationships are routed as well as what they connect.
Orthogonal routing + obstacle avoidance
ELK places the graph; libavoid reroutes eligible candidates around table obstacles.
Shared buses + high-fanout junctions
Related references can share a trunk. Select an individual relationship without losing its original endpoints.
Four-sided connection ports
Ports can use all four table sides while retaining field identity and 1/N cardinality.
This is ER Wiki’s own output, not a fabricated before/after comparison. The 13-table example demonstrates the interaction, not a large-schema benchmark.
Read the whole schema. Inspect one relationship.
Navigate by scope
Overview → domain → table → column. Search, a minimap and focused tracing help keep your place.
Read the meaning
Inspect fields, enums, cardinality and recorded evidence. A drawn logical relationship is not automatically a database foreign key.
Work on your model
Desktop: import SQL locally, edit tables and relations, round-trip JSON, export SVG/PNG. The web demo is read-only.
Local work, without a hosted account
The installed Electron app works offline with local models. No application telemetry or automatic model uploads. Update checks contact GitHub only when requested; the hosted demo still uses normal web requests.
Built on existing tools, not a new diagram engine
drawDB provides the editing foundation. ELK handles layout and compaction; libavoid handles obstacle routing. React renders the workspace, and Electron packages the desktop app.
Read the implementation →Bring a schema that no longer fits on a whiteboard.
Start with the fictional example. Use the desktop app for your own SQL and JSON models.
Published binary: macOS Apple Silicon. Ad-hoc signed, not notarized. Windows/Linux releases are not validated.