Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore how S3 RDMA technology revolutionizes object storage performance for AI/ML workloads in this 47-minute conference talk from SNIA SDC 2025. Learn about the limitations of traditional Amazon S3 access via TCP/IP, which is CPU-intensive and inadequate for the low-latency, high-throughput demands of modern GPU workloads. Discover how S3 RDMA implements S3 object PUT/GET data transfers over RDMA, completely bypassing the HTTP stack and enabling zero-copy data flow directly into user memory space. Examine Dell's practical experience building an S3 RDMA solution for Dell ObjectScale, including valuable insights into client-side data path efficiency improvements, optimized GPU utilization, and high-speed data transfer with reduced control messages. Compare the benefits of S3 RDMA against traditional S3 access methods and other RDMA-based storage technologies. Understand the specific advantages S3 RDMA provides for AI/ML workloads and its potential for interoperability and standardization in cloud storage. Gain insights into emerging trends in cloud storage and AI/ML data access patterns from Dell Technologies' Jason Goldschmidt.
