zingdev — infra

The diagram here is a representation of the infrastructure I run all of my side projects on. It is a self managed k3s cluster running on 5 Hetzner Cloud VMs. It is bootstrapped through IAC and most of the Kubernetes infrastructure is managed via FluxCD. Go ahead, click and drag around.

drag, scroll and click the diagram to explore.

The slider below uses a websocket in your browser to send traffic to my cluster. Live metrics from the cluster are reported in the graph below. Once you increase the load past where the current pods can keep up, a backlog of events starts to grow. The cluster monitors this backlog and scales workers until the backlog is handled. The architecture diagram in the background displays the flow of data and will update as new workers go on/offline.

offered TPSsystem TPSbacklogworkers

load0 TPS

running · auto-stops in 0s

offline

global cap (500 TPS) reached — load shed.

0
system TPS: 0
backlog: 0
worker pods