Hosting small to medium-sized Drupal sites isn't typically a big challenge. But what about large sites where up-time is business-critical for a customer?

How should we as Drupal sys admins best manage high-traffic servers? — servers required to handle thousands of users simultaneously (like our recently launched "The Voice of Switzerland" website). The process of simply duplicating the web server isn't enough; issues will undoubtedly occur and persist.

Foregoing sleep is of course one option, but an unattractive one (and certainly not the solution for me).

Building a high performance site is more than only setting up and configuring the correct server software. There's a lot of planning that needs to happen in the beginning. For example, while themers might enjoy using node teaser displays, those very displays can be deathly for a server.

At Amazee Labs we've set up our Drupal cluster with multiple web servers, Varnish, Memcache, MySQL Replication, GlusterFS, and more. During this presentation we'll walk through these solutions and also cover some special configurations for Varnish to cache AJAX calls, a feature which is not usually cached.