tscolari / bender

Benchmark tool/library

Bender

Bender is a tool to benchmark external commands executions.

USAGE:
   main [global options] command [command options] [arguments...]

GLOBAL OPTIONS:
   --count value        how many times should the command run (default: 1)
   --concurrency value  how many threads to use (default: 1)
   --command value      command(s) to run. May be set more than once
   --keep-running       run until aborted (ctrl-c)
   --interval value     interval to use between each call when using keep-running (default: 0s)

Output

The output is in JSON format:

$ bender --count 5 --command ls --command "sleep 1" --command "sleep 3" --concurrency 3
{
  "commands": {
    "1": {"exec":"ls","run_count":2},
    "2": {"exec":"sleep 1","run_count":2},
    "3": {"exec":"sleep 3","run_count":1}
  },
  "duration": 3000814674,
  "success_counter": 5,
  "error_counter": 0,
  "each_run":[
    {"command":1, "duration": 717232, "start_time": "2017-04-24T21:23:08.830283485+01:00", "failed": false},
    {"command":1, "duration": 704930, "start_time": "2017-04-24T21:23:08.830326133+01:00", "failed": false},
    {"command":2, "duration": 1000535314, "start_time": "2017-04-24T21:23:08.831003851+01:00", "failed": false},
    {"command":2, "duration": 1000533965, "start_time": "2017-04-24T21:23:08.831032164+01:00", "failed": false},
    {"command":3, "duration": 3000735861, "start_time": "2017-04-24T21:23:08.83033188+01:00", "failed": false}
  ]
}

• commands: is an indexed list of all the commands that were passed as arguments, with the total run count of each.
• duration: is the duration of the execution
• success_counter: number of commands that did not exit in error
• error_counter: number of commands that did exit in error
• each_run: a summary of each command run containing:
  • command: the index of the command from the commads key
  • duration: duration of that execution
  • start_time: when the execution started
  • failed: true if the execution exited in error

Installation

go get github.com/tscolari/bender

or download from releases

Count

When using --count it will run the commands defined by --command for a finite amount of times. When finished it will summarize the results

KeepRunning / Interval

When using --keep-running the benchmark will run forever, or until it receives a signal to terminate. It's possible to set --interval to force the benchmark wait between each run. Please note that if --concurrency is set to anything bigger than one, the interval will apply for each thread individually.