Running Databases Efficiently at Scale - Case Study of Flipkart's Local PV Journey on K8s

Learn how to run large-scale, performance-sensitive distributed databases on Kubernetes through this 50-minute conference talk from KubeCon + CloudNativeCon. Discover Flipkart's practical approach to leveraging Kubernetes Local Persistent Volumes (Local PVs) to optimize both performance and infrastructure costs for stateful workloads. Explore the technical challenges and solutions involved in managing a massive storage infrastructure comprising 4PB of directly attached SSDs and 15PB of directly attached HDDs across 1000 Kubernetes stateful nodes spanning three clusters. Gain insights into how Local PV storage powers various internal database platforms including Aerospike, HBase, and TiDB that support critical business applications. Examine key implementation aspects including dynamic provisioning strategies, workload scheduling optimization, and automated maintenance workflows designed to ensure operational efficiency at enterprise scale. Understand the balance between latency, reliability, and cost efficiency when deploying distributed databases in cloud-native environments.