Overview
Watch a technical presentation from Google Cloud's storage group exploring advanced storage solutions optimized for AI and ML workloads. Learn about key innovations like Anywhere Cache for accelerated GPU and TPU access, Intel DAOS-based parallel file systems for ultra-low latency, and HyperDisk ML enabling read-only data access across thousands of hosts. Discover how Cloud Storage FUSE with caching allows bucket mounting as a file system, reducing storage costs and improving training efficiency. Explore Filestore's various instance types (Basic, Zonal, and Regional) designed for different performance needs, along with features like Multi-Share for automatic scaling and HyperDisk storage pools for optimized resource management. Gain insights into Google Cloud's comprehensive approach to supporting both AI/ML and enterprise workloads with high availability and seamless failover capabilities.
Syllabus
AI/ML Storage Workloads in Google Cloud
Taught by
Tech Field Day