How can I run multiple functions in k6 simultaneously, with specified rate per function

Question

I'm trying to figure out a way to issue two separate requests in k6, with each request limited in relation to the other.

For example, I have two POST requests created, I want one to be hit 90% of the test run and the other 10%, to simulate a more busy and less busy endpoints being hit by requests on the same HTTP server.

EDIT: The point of the exercise is not to generate a predictable load, but to generate a maximum (in terms of ops/sec) load the server can take and measure it. The different rates of request types are needed for the benchmark to simulate a realistic load.

Answer 1

Use scenarios to call different test functions with different request rates by specifying a constant arrival rate executor.

You have to pre-calculate the per-endpoint rates manually, but you could easily define a function to compute the correct rates from a total base rate.

const TOTAL_RPS = 100;
function rate(percentage) {
  return TOTAL_RPS * percentage;
}

export const options = {
  scenarios: {
    busy: {
      exec: 'busy_endpoint',
      executor: 'constant-arrival-rate',
      duration: '10m',
      preAllocatedVUs: 50,
      rate: rate(0.9) // 90%
    },
    lazy: {
      exec: 'less_busy_endpoint',
      executor: 'constant-arrival-rate',
      duration: '10m',
      preAllocatedVUs: 5,
      rate: rate(0.1), // // 10%
    }
  }
};

export function busy_endpoint() {
  // will be called 90 times per second (90% of 100 rps)
  http.get('http://example.com');
}

export function less_busy_endpoint() {
  // will be called 10 times per second (10% of 100 rps)
  http.get('http://example.com');
}

---

Depending on your requirements and the complexity of the logic, an alternative could be a stochastic approach: each time you enter the method, you generate a random number and based on the result take one or the other endpoint (or any logic for that matter).

What could that look like?

function busy_endpoint() {
  // will be called roughly for 90% of your requests
  http.get('http://example.com');
}

function less_busy_endpoint() {
  // will be called roughly for 10% of your requests
  http.get('http://example.com');
}

function get_endpoint() {
  const r = Math.random();
  if (r < 0.1) return less_busy_endpoint;
  return busy_endpoint;
}

export default function() {
  const endpoint = get_endpoint(); // gets busy with 90% chance and less_busy with 10% chance
  endpoint(); // calls the endpoint
}