Request time

This wiki page explains the design of the feature that returns in an HTTP header the time every request took to be answered.

Context

Solution

Every HTTP request goes through the platform's API Gateway, which alows the counting of the time taken by each request.

This is achieved by using an Instrumentation (ruby) rack middleware that simply collects the time when the request entered the middleware, calculating the difference when the request returns, and storing it in the X-Timing HTTP header of the response. In code (with irrelevant code ommited), we have:

  ...
  def call(env)
    began_at = Time.now
    status, headers, body = @app.call env

    headers['X-Timing'] = (Time.now - began_at).to_f.to_s
    [status, headers, body]
  end
  ...