ZaloPay Cannon

Introduction

ZaloPay Cannon is a benchmark system for ZaloPay's internal services. The aim is to build a multi-tenant system which provides intuitive UI/UX for users to submit tasks and perform benchmark.

Architecture

Features

• Benchmark gRPC service with given proto.
• Distributed testing: run tests on multiple slaves.
• Automatically generate input data.
• Visualize data: metrics data is stored and visualized using InfluxDB.
• Only support Unary RPCs.

Requirements

• Golang 1.13.1
• Locust
• Influxdb 2.0
• Python 3.7.3

Usage

• Make sure Locust, InfluxDB and gRPC server are running.

Usage:
  cannon run [flags]

Flags:
  -c, --config string        path of config file
  -r, --hatchRate int        config hatch rate (users spawned/second) (default 10)
  -h, --help                 help for run
  -H, --host string          target gRPC host (default "localhost")
      --locust-host string   host of locust master (default "localhost")
      --locust-port int      port of locust master (default 5557)
      --locust-web string    locust web target (default "http://localhost:8089")
  -m, --method string        method name
  -w, --no-workers int       number of workers to simulate (default 10)
  -P, --port int             target gRPC port (default 8000)
  -p, --proto string         path of proto file

• Cannon example config file

# Locust Config
LocustWebPort: "http://0.0.0.0:7000/"
LocustHost: "127.0.0.1"
LocustPort: 5557
NoWorkers: 80 # Number of connections
HatchRate: 10 # Hatch rate

# InfluxDB Config
IsPersistent: "true" # turn on if you want save metrics in influxDB
Bucket: "benchmark-results" # InfluxDB bucket's name
Origin: "zlp-osss"
DatabaseAddr: "http://0.0.0.0:9999"
Token: "egc6_K6V0pCmEwIahIzmnoneommTcsa7TS5XtmcSBnR9VeX31dMsRJ_STN-bUqOwWW77vPiU0aM9RGMQFwxT-A=="

# gRPC benchmark target
GRPCHost: "localhost"
GRPCPort: 4770
Proto: "./proto-name.proto"
Method: "serviceName.methodName"

• You can start multiple instances of cannon to benchmark in distributed mode.
• Monitor at dashboard locust

Example

Read the example

Roadmap

Read the roadmap

Acknowledgements

• Thanks to @anhld2, which served as an inspiration and guide in building this project.
• Special thanks to @thinhda, @tranndc and @quyenpt3 for their work on making component-based theming a reality.