HTTP Connection Management, From 10 Users To 100 million

Operations and Culture
Location: Hall 1 B/C Level:
Average rating: ****.
(4.46, 41 ratings)

A website’s infrastructure always starts simply and for good reason, we don’t have users and we want to get back to writing code.

But, as a website grows, that Apache config you copied from stackoverflow is going to hit the wall. So in this talk we’ll cover the problems, signs of impending doom and typical solutions needed to get from 10 users to 100 million.

Bullet points!

  • Common web servers and how far they can get on their own.
  • Signs of trouble, or why ‘top’ won’t tell you there’s a problem.
  • Load-balancing, software vs hardware.
  • Making the web faster for your users, especially in Europe.
  • Oh, load-balancing webservers, that’s quaint. Try load-balancing continents. Love, GSLB.

This is a mid-level technical talk, aimed at the developer or ops guy who’s looking at the next step, or wondering where the last one went wrong.

Photo of Bradley Heilbrun

Bradley Heilbrun

YouTube/Google Inc

YouTube’s first ops engineer. Currently a Google SRE based in Paris and still serving cat videos.

Comments on this page are now closed.


Ian McDowall
10/11/2011 17:47 CET

Really cool, do you plan to post the slides?

Abdul Karim
9/11/2011 20:35 CET

really interesting and insite into youtube.



  • ip-label
  • Compuware Corporation
  • dynaTrace
  • Keynote Systems
  • New Relic
  • Citrix Systems
  • Google
  • Apica
  • AppDynamics
  • CDNetworks
  • Cotendo
  • Dyn Inc.
  • ImmobilienScout24
  • Spil Games
  • Dyn Inc.

For information on exhibition and sponsorship opportunities at the conference, contact Gloria Lombardo at

View a list of Velocity Europe contacts