At my work (large corporate office), we have random node outages. It's not quite as in depth as chaos monkey, but it goes towards the same purpose. Just pull the plug on the server. More than once, a random node outage has caught a novice developer making static links to nodes through the load balancer. We also have random pen-tests designed to DoS or otherwise disable services around the network. Controlled destruction of your infrastructure is the quickest way to highlight any faults.

But remember: what's the difference between hacking and pentesting? Permission.