Pinterest was born on AWS. Today, we operate thousands of instances and process over 150PB of data.
In this talk, I'll share why & how Pinterest moved from VMs to Containers and K8S. 1) Pinterest Infrastructure Overview 2) VMs vs. Containers a. Developer Velocity - We will discuss the overall job lifecycle workflow i.e build, setup, deploy, operations when using VMs and Containers. b. Service Reliability - Constraints around resource isolation c. Infrastructure Governance - Attribution of resources both on utilization & Spend d. Efficiency – Specifically around auto-scaling. We use ASGs heavily and will share how we quantified efficiency when in VM vs. Container 3. Move to Containers - I'll share how Pinterest adopted Docker & subsequently evaluated different orchestration systems across dimensions such as docker integration, scheduling, security, networking, etc.