Flash Memory Summit (FMS) 2020 – NVM Express Presentation Track

This session discusses the real world capabilities of the NVMe 1.4 and NVMe-oF 1.1 specifications. Viewers will learn about the applications Facebook’s hyperscale data center uses, and how they benefit from NVMe features in deploying at scale. Viewers will also learn from HPE on how NVMe over Fabrics is evolving to redefine next generation SAN and the key fabric requirements to enable this new frontier.

The future surely belongs to speed and data-hungry applications such as AI, real-time analytics, VR/AR, high-resolution video, amazing games, 5G and IoT. NVM Express (NVMe) technology offers a proven road to making them widely available at low cost. Recent NVMe architecture advances such as zoned namespaces, persistent memory regions, and higher performance interfaces such as those provided by PCIe 4.0 and 5.0 technologies are just the beginning. NVMe technology is being extended to support Hard Disk Drives (HDDs) and computational storage.  Computational storage will enable a new level of storage performance by moving the execution of programs that operate on data to NVMe storage devices, increasing parallelism and eliminating the need to transfer vast amounts of data to CPU memory. Join us in developing the storage architecture that will power the connected universe.

NVMe is the fastest growing storage technology of the last decade and has succeeded in unifying client, hyperscale, and enterprise applications into a common storage framework. NVMe has evolved from a being a disruptive technology to becoming a core element in storage architectures. In this session we will talk about the NVMe transition to a merged base specification inclusive of both NVMe and NVMe-oF architectures. In this presentation, Nick Adams provides an overview of the latest NVMe technologies, summarize the NVMe standards roadmap, and describe the latest NVMe standardization initiatives. There are a number of areas of innovation that preserve our simple, fast, scalable paradigm while extending the broad appeal of NVMe technology. These continued innovations will ready the NVMe ecosystem for yet another period of growth and expansion.

On a regular cadence, the NVM Express Interoperability and Compliance Program is updated to conform with the latest NVMe specifications. This presentation will provide updates on the latest community efforts and program updates that companies need to know before submitting products for testing. In this presentation, David Woolf outlines the recent updates to the official NVM Express Compliance program, including how agnostic to underlying transport enables smooth transition from PCIe 3.0 to 4.0 to 5.0 specifications. Additionally, we will cover how driver development for both SSDs and NVMe-oF architecture, especially in Linux, enables adoption. Finally, updates will include Connector development for key use cases and a description of how different media types (NAND, Optane, HDD, other Flash types) are masked by NVMe technology interoperability.

Agenda:

• NVMe Technology at Scale – Lee Prewitt
• NVMe Technology and Flash SSDs in Cloud Applications – Kamaljit Singh
• NVIDIA NVMe® Technology in the Cloud – John Kim
• Deploying NVMe Flash at Facebook – A Journey – Wei Zhang

Exploiting the full SSD performance in scalable disaggregated architectures is a continuous challenge. NVMe/TCP, released in 2018, enables a broader sharing of distributed storage resources. It complements NVME-oF over RDMA, avoiding performance degradation over distant links and simplifying the deployment. However this comes at the cost of an heaviest networking stack and requires the latest Linux kernels. In this talk, Sagi Grimberg analyzes the differences between RoCE and TCP, and describes how to eliminate bottlenecks, achieving best-in-class performance for both protocols in an end-to-end NVMe-oF communication.