Mark is a passionate engineer who loves to solve hard problems. He is an infrastructure engineer at Plaid, a financial data network, where he works at solving scaling issues around deploying software, and keeping the infrastructure running. He specifically enjoys focusing on CICD pipelines, metrics and monitoring, and developer optimization. Mark attended the University of Victoria where he developed a microservices-based submarine control and navigation system, and received a Bachelor of Engineering. His passion for learning new things has led him to work in many business domains over his 15 year career. He has previously developed GIS mapping systems, pension management systems, source code static analysis tooling, and car rental software. In his spare time, Mark rehabilitates under-socialized kittens, renovates his heritage house, and makes Jeff Bezos cry by reducing AWS spend.
Zero-touch deployments: keeping 300 devs from needing to learn Kubernetes
Plaid recently found themselves in a position where; they were growing fast and needed to find ways to scale their engineering operations. While the process for submitting code was well-established, there were several manual steps required to move that code into production. The result was that code wasn’t deployed promptly, resulting in fewer deploys with dozens of commits each. To solve this problem, Plaid built a fully automated deployment system that could monitor and rollback services if they failed advanced health checks. This meant that engineers could focus on writing high-quality code with highly-reliable deploys. This talk will cover how Plaid built this system using open-source tools like Kubernetes, Argo Rollouts, and LinkerD, and discuss both the advantages and drawbacks.