Evolve or Die: High-Availability Design Principles Drawn from Google's Network Infrastructure

   Abstract