CODEWITHSUNDEEP
concurrency
stk1234
stk1234

Reputation: 1116

Concurrent Requests for an API

Is it possible to roughly estimate how many concurrent requests an API might receive?

Let's say it's a super simple API that just returns "hello" to a GET request deployed on a 16gb machine. In general, how many concurrent requests could it support before it starts to melt or say nah?

If it failed because of too many concurrent requests, what would happen?

Upvotes: 0

Views: 364

Answers (1)

stk1234
stk1234

Reputation: 1116

As PiRocks suggested, I ran an experiment

  1. Deployed a simple node.js api app to heroku
  2. Deployed the app to heroku (machine specs TBD - looking around if they even list it)
  3. Signed up for a free account on loader.io

Unfortunately, the maximum for free is 10k requests over 15s, aka 666 QPS. That resulted in a 2ms average response time, no timeouts, and no errors. Might upgrade to see what it looks like from there.

Update: seems like 2K QPS is where I started to see errors. More details here

Upvotes: 1

Related Questions