Paper Title: Advantages of Ethernet SSDs in AI Applications

Paper Abstract: The emerging new technology of Ethernet SSDs offers a unique combination of lower cost and high performance. These two value propositions seldom come as a pair with a new technology. For Al applications, the focus is on the performance, and no longer fits in a single GPU server's local storage, and remote storage is needed. These applications need storage with extremely low latency, but with high bandwidth. Today's remote block storage JBOF controller architectures are fundamentally store and forward. CPU complexes convert local NVMe protocol to NVMe over Fabrics and visa versa. This adds latency and potential bottlenecks at the memory bus and network ports. Learn how Ethernet SSDs (ESSD) in EBOFs (Ethernet Bunch of Flash) remove those performance barriers allowing the performance capabilities of the SSD to show through to the GPU and its AI applications.

Paper Author: Rob Davis, VP Storage Technology - Networking Business Unit, NVIDIA

Author Bio: Rob Davis is the VP of Storage Technology at the NVIDIA Networking Platform Group enabling ways to apply high-speed interface technology to storage systems. For more than 35 years, he has been a key contributor and decision maker for the development of storage networking products, including products based on NVMe over Fabrics technology. Rob was previously VP and CTO Officer at QLogic, where for 15 years he drove development and marketing of Fibre Channel, Ethernet and InfiniBand technologies. His in-depth expertise spans virtualization, Ethernet, Fibre Channel, SCSI, iSCSI, InfiniBand, RoCE, SAS, PCI, PM, SATA, and flash Storage.