Back to SamiBeats

A Network That Refuses to Die

Break a network. Watch it route around damage in real time. Let AI analyze risks and make recovery decisions.

What's Happening
The network is operating normally.
Data packets travel from source nodes (blue) to destination nodes (green) along the shortest available paths. Every node and connection is healthy.
Technical Details ▼
Routing: Dijkstra's shortest-path algorithm with weighted costs per node status and edge latency.
Recovery: Degraded nodes start recovering after 5s, full health restored after 3s more. Connected edges also restore.
Rerouting: When a node fails or edge is cut, all active packets instantly recalculate their routes using the current graph state.
Health Score: Weighted composite: 60% delivery rate + 20% connectivity + 10% redundancy + 10% latency.
Topology: Range-based mesh. Each node connects to all peers within radio range. Resilience level controls range (fragile=1.6x, balanced=2.0x, robust=2.7x grid spacing). Latency is proportional to distance.
Chaos Types: Kill Node (removes node + all edges), Cut Connection (severs one edge), Slow Zone (degrades a cluster). Click any node to manually toggle it.
AI Analysis: Betweenness centrality identifies critical routing nodes. Tarjan's algorithm finds articulation points (single points of failure). Recovery priority uses route impact scoring.