Skip to content

Latest commit

 

History

History
62 lines (38 loc) · 2.06 KB

README.md

File metadata and controls

62 lines (38 loc) · 2.06 KB
Raw

Summary

CF application to run bench scenarios on external services. Each service provides 2 scenarios, nominal & benchmark.

Nominal scenarios will launch a light test in order to check that your service works properly.

Benchmark scenarios will launch a heavy test to benchmark your service.

Currently supported services :

  • mysql using sysbench as a benchmark tool
  • redis using redis-benchmark as a benchmark tool

Currently implemented scenarios :

  • mysql : nominal
  • redis : nominal

⚠️ Please be advised that benchmarks need their own database, don't use a production database, it will result in dataloss. ⚠️

Installation

pre-requisites

  • This application first needs a redis service in order to store result and queue awaiting benchmarks. This service name MUST starts with benchmark-redis-storage

    cf create-service redis plan benchmark-redis-storage

  • Other services you will bind your application to will be benched, be careful as bench may resul in data loss

  • You need to have pip3 installed to download vendor packages

  • You need to be logged in your org/space and ready to cf push

configuration

  • Open deploy_on_cf.sh
  • Fill APP_REDIS_STORAGE with redis service you previously created
  • Fill APP_SERVICES_TO_BENCH variable with services you wish to bench. If you want to bench multiple services, separate them with a space
  • run ./deploy_on_cf.sh
  • Optionnal : setting DONT_USE_REDIS_BENCHMARK will use a redis-benchmark substitute to avoid firewall issues

Usage

API exposes 2 routes :

  • /run : will add an entry in redis queue to start benchmarks.
  • /results : returns JSON object containing benchmarks results
  • /metrics (to be implemented) : returns results in prometheus format

Just curl api_url/run to start benchmarks, results will appear on api_url/results as soon as they are available.

If you run a new bench, it will create new results, and so on.

Architecture

API

API uses Flask & Celery to provide API and queuing

worker

Worker uses Celery to run tasks