cagraham writes: While most tech companies have extensive backup plans in place in case of server failure, Netflix takes things one step further and actually sabotages its own infrastructure. According to their cloud director Ariel Tseitlin, Netflix releases code that randomly causes production instances to terminate. They do this every workday sometime between 9-5, in order to expose their own vulnerabilities. They do it during the workday so that their engineers don't have to come back in after-hours to fix the problems. While these tests are probably stressful, apparently Netflix's infrastructure is built around isolation and redundancy in order to minimize just this kind of damage.