/flexible-raft-benchmark

Flexible Raft benchmark for TiKV

Flexible raft changes - https://github.com/akashsharma95/raft-rs/tree/captain/flexible

Create TiKV cluster

Before creating cluster resources delete any existing ones

Steps to delete any existing resources created for benchmark

  • Delete NetworkChaos kubectl delete networkchaos network-chaos-delay -n chaos-testing

  • Check for existing TiKV cluster kubectl get tikvcluster -n default

  • Delete TiKV cluster kubectl delete tikvcluster <cluster_name> -n default

  • Check for existing PVC (block storage) kubectl get pvc --selector=app.kubernetes.io/instance=test-cluster -n default

  • Delete existing PVCs kubectl delete pvc --selector=app.kubernetes.io/instance=test-cluster -n default

Steps to recreate cluster

  • Apply TiKV cluster manifest kubectl apply -f ./ycsb/tikv-cluster-baseline.yaml

  • This will launch PD pods and TiKV pods as a statefulsets with PVC bound to it Check if all the resources are up kubectl get pods --selector=app.kubernetes.io/instance=test-cluster

  • Apply network chaos to simulate network latencies kubectl apply -f ./ycsb/network-chaos-delay.yaml .spec.value defines how much pods you want to add delay on

Run YCSB Benchmark

  • Load BenchmarkF ./bin/go-ycsb load tikv \ -P $GOPATH/src/github.com/pingcap/go-ycsb/workloads/workloadf \ -p tikv.pd="test-cluster-pd:2379" -p tikv.type="raw" \ -p tikv.conncount=64 -p tikv.batchsize=64 \ -p operationcount=100000 -p recordcount=100000 \ -p threadcount=100

  • Run BenchmarkF ./bin/go-ycsb run tikv \ -P $GOPATH/src/github.com/pingcap/go-ycsb/workloads/workloadf \ -p tikv.pd="test-cluster-pd:2379" -p tikv.type="raw" \ -p operationcount=100000 -p recordcount=100000 \ -p threadcount=100