Speed Racer - Local Persistent Volumes in Production

Explore the intricacies of Local Persistent Volumes in Kubernetes production environments in this 25-minute conference talk by Matt Schallert from Chronosphere. Gain insights into leveraging this powerful feature for building high-performance stateful systems while navigating potential pitfalls. Learn about in-cluster volume provisioning, best practices not readily apparent in Kubernetes documentation, and cloud provider-specific local disk behavior. Discover concrete tips for avoiding common mistakes, day-two operational best practices, and strategies for handling node failures, backups, and cluster operations. Understand the tradeoffs of local volumes, their limitations, and ideal use cases. Walk away with a comprehensive understanding of how to effectively implement Local Persistent Volumes in your Kubernetes workloads, including anti-patterns to avoid and better alternatives like replicated data systems.

Intro
Local Disks
Local Disk Performance
Limitations of Persistent Volumes
How Local Volumes Work
Day Two Operations
Node Failures
Backup & Restore
Cluster Operations
Node Upgrades: Local Volumes
Anti-Pattern: Single Primary
Better Fit: Replicated Data
Use Cases
Summary
Resources