In this episode of the Prodcast, Cody Smith and Trisha Weir share their journey of rebuilding Google's front-end infrastructure, known as Maglev. Faced with the limitations and high costs of their vendor's network load balancers, they initiated a skunkworks project. By tapping into their internal expertise and fostering a high-trust environment, they quickly developed and launched a new system. The rollout was executed through a gradual migration, focusing on VIPs, which underscored the significance of iterative development, careful monitoring, and adapting to the complexities of the system. Their conversation also highlights the importance of generalist skills, psychological safety, and teamwork within Site Reliability Engineering (SRE) teams.
