TSM-GNS-APP-Perf

Prepare Cluster

prepare cluster
Install Istio
install prometheus step by step https://devopscube.com/setup-prometheus-monitoring-on-kubernetes/
git clone https://github.com/istio/tools.git

kubernetes create ns fortioclient
kubernetes create ns fortioserver
kubernetes -n fortioclient apply -f twopods/fortio_test_client.yaml
kubernetes -n fortioserver apply -f twopods/fortio_test_server.yaml

Prepare Python Environment

cd tools/perf/benchmark
cd runner
pipenv shell
pipenv install
cd ..

Run performance tests

export NAMESPACE=fortioclient
export DNS_DOMAIN=local
python3 ./runner/runner.py --conn 1,2,4,8,16,32,64 --qps 1000 --duration 240 --baseline --telemetry_mode telemetryv2
python3 ./runner/runner.py --conn 16 --qps 100,200,400,800,1000 --duration 240 --baseline --telemetry_mode telemetryv2

Gather Result Metrics

kubernetes -n monitoring port-forward prometheus-deployment-7bb6c5d7fd-mb9dm 9090:9090 &
kubernetes get svc -n fortioclient fortioclient // get external_ip
export FORTIO_CLIENT_URL=http://external_ip:8080
export PROMETHEUS_URL=http://localhost:9090
python3 ./runner/fortio.py $FORTIO_CLIENT_URL --prometheus=$PROMETHEUS_URL --csv StartTime,ActualDuration,Labels,NumThreads,ActualQPS,p50,p90,p99,cpu_mili_avg_telemetry_mixer,cpu_mili_max_telemetry_mixer,mem_MB_max_telemetry_mixer,cpu_mili_avg_fortioserver_deployment_proxy,cpu_mili_max_fortioserver_deployment_proxy,mem_MB_max_fortioserver_deployment_proxy,cpu_mili_avg_ingressgateway_proxy,cpu_mili_max_ingressgateway_proxy,mem_MB_max_ingressgateway_proxy

Example of Echo/Http/Grpc Ping

kubernetes -n fortioclient exec fortioclient-POD -- fortio curl http://fortioserver.gns-sc.local:8080
kubernetes -n fortioclient exec fortioclient-POD -- fortio load -a -grpc -ping -grpc-ping-delay 0.25s -payload "01234567890" -c 2 -s 4 -labels grpc http://fortioserver.gns-sc.local:8076
kubernetes -n fortioclient exec fortioclient-POD -- fortio load -jitter=False -c 2 -qps 1000 -t 93s -a -r 0.001 -httpbufferkb=128 -labels bb http://fortioserver.gns-sc.local:8077/echo?size=1024

Visualize Results

cd graph_plotter/
python3 ./graph_plotter.py --graph_type=latency-p999 --x_axis=conn --telemetry_modes=telemetryv2_both,telemetryv2_baseline --query_list=1,2,4,8,16,32,64 --query_str=ActualQPS==1000 --csv_filepath=250PODS_SC.csv --graph_title=p999.png

python3 ./graph_plotter.py --graph_type=ingressgateway-cpu --x_axis=qps --telemetry_modes=telemetryv2_both,telemetryv2_baseline --query_list=10,100,200,400,800,1000 --query_str=NumThreads==16 --csv_filepath=250PODS_SC.csv --graph_title=cpu.png