Latency Headers PoC

Adding HTTP headers to http responses enables developers to gain a better view into their API layer's latency and networking bottlenecks.

Explaination

Basically there are two headers introduced in this proof of concept:

• x-request-received is set by the server with the timestamp of when the request was received
• x-response-sent is set by the server with the timestamp of when the response was sent

With these headers in place the client can then determine the following:

• outgoing network latency: time between client sending the request and server receiving it
• server processing latency: time between server receiving the request and sending the response
• incoming network latency: time between server sending the response and client receiving it
• total round trip latency: time between client sending the request and receiving response

Complete Example

server.js

// Load the http module
var http = require('http');

// Make it easy to pkill from npm
process.title = "latencyServer"

// Create the server
var server = http.createServer(function (request, response) {

  // Set the x-request-received header with the current timestamp
  response.setHeader('x-request-received', new Date().getTime());

  // Simulate a delay for processing latency to show up
  setTimeout(function(){

    // Set the x-response-sent header with the current timestamp
    response.setHeader('x-response-sent', new Date().getTime());

    // Set the header status to ok and the content type
    response.writeHead(200, {"Content-Type": "text/plain"});

    // Return the demobligatory hello world response
    response.end("Hello World\n");

  // Remember that delay? 50-75ms sounds like a good target
  }, Math.floor((Math.random() * 25) + 50))

});

// Listen on localhost:1337
server.listen(1337);

app.js

// Require http for requests
var http = require('http')

// Save the timestamp of when the request was sent as its required
var requestSent = new Date().getTime()

// Send the request to a server that returns the latency headers
http.get("http://localhost:1337", function(res) {

  // Save the response headers for easy access later
  var headers = res.headers

  // Save the timestamp of when the response was received
  var responseReceived = new Date().getTime()

  // Save the headers we care about in variables
  var responseSent = headers['x-response-sent'] || false
  var requestReceived = headers['x-request-received'] || false

  // If the latency headers do not exist
  if (!responseSent || !requestReceived){
    throw new Error("The server did not respond with latency headers.")
  }

  // The math to determine the latencies
  var outgoingLatency = requestReceived - requestSent
  var processingLatency = responseSent - requestReceived
  var incomingLatency = responseReceived - responseSent
  var roundtripLatency = outgoingLatency + processingLatency + incomingLatency

  // Return back the latencies
  var results = {
      "outgoingLatency" : outgoingLatency,
      "processingLatency" : processingLatency,
      "incomingLatency" : incomingLatency,
      "roundtripLatency" : roundtripLatency
  }

  // Output the results
  console.log(results)
})

Try it out

An example server and app is included in this repo for your convenience.

# Clone this repo
git clone [email protected]:montanaflynn/latency-header-benchmark.git

# CD to the example dir
cd latency-header-benchmark/example

# Install this package (npmception?)
npm install

# Start the server in the background
node server.js &

# Run the app
node app.js

# Kill the example server
pkill latencyServer

Related projects

• Latency Header Benchmarker
• Express Latency Header Middleware
• Koa Latency Header Middleware

Copyright (c) 2014, Montana Flynn (http://anonfunction.com/)