Simiload

Simulation of different load shedding strategies. The high level architecture is as follows:

Running

Simulation server:

make compiles server.go and runs a container, within the metrics cluster network, with the simulation server at: localhost:8080

Metrics cluster:

make metrics starts a metrics collection cluster, the Grafana frontend is at: localhost:3000

Load Generation:

To generate some load use: go run generate.go -config flash_sale.json "http://localhost:8080"

Dashboard:

Configure a Grafana data source of type prometheus. The API URL is http://simiload_prometheus_1:9090
Import the dashboard stored in dashboard.json file in the repo

I used the logs for quick development:

And metrics for tuning: