Surge 2012 ~ Andrew Miklas ~ Ensuring the Call Goes Out - Everytime
Many systems can afford a bit of downtime now and then. Unfortunately, at PagerDuty, even a three-minute outage at the wrong time can cost thousands in lost revenue. This talk will cover some of the techniques and tricks we use at PagerDuty to ensure that our phone and SMS alerts continue to flow while staying well within the budget of a typical startup. Specifically, I will show how we do zero downtime database migrations and deploys, and how we recover from host and data center outages with minimal disruption.