Surge 2012 ~ Joe Kottke ~ Scaling in the Cloud at Cost and SLA
I will cover the techniques my team has used to achieve a 99.995% uptime using a service-oriented infrastructure in a cross-cloud heterogeneous environment. Highlights will include local and geographic load-balancing, elastic scaling for dynamic traffic patterns, and using polyglot persistence techniques for near-real-time data distribution across geographic locations. I will also cover the creative ways in which simple distributed system techniques were used to avoid single-points-of-failure, as well as deployment tricks that allow the system to be updated with zero down-time; all with a team of only 12 engineers and open-source software.