NVMe Over Fabric is the latest buzz in storage industry. Almost all the storage companies are developing their storage appliances with NVMeOF support and few of the companies has already launched their product with basic storage functionality support. During this presentation, I will open my presentation with key challenges customers are facing in adopting NVMeOF based solutions and where are Industry today to solve these challenges.

NVMe protocol is optimized for NAND media, providing high performance in comparison with legacy protocols, such as SAS and SATA. Furthermore, it replaces the conventional volumes with namespaces and subsystems. NVMe-oF protocol enables sharing the NVMe resources within the rack and across racks in the datacenter, while maintaining the benefits of NVMe. In our talk, we will describe how NVMe SW abstraction further increases the potential of NVMe-oF. The abstraction enhances the ability to share storage resources.

Enterprise storage are using advanced storage controller, flash based devices to improve performance along with advanced protocols like NVMe for Direct attached storage. We are getting advanced on storage front but we are not getting the enough performance within the data center as there has been a bottleneck at network level. There was a need to implement a network which can retain flash speed from Host (Server) to device (Storage). Latest technology disturbance in storage industry is NVMe Over Fabric, which takes the low latency of NVMe protocol over network.

Network-intensive applications, like networked storage or clustered computing, require a network infrastructure with high bandwidth and low latency. Remote Direct Memory Access (RDMA) supports zero-copy data transfers by enabling movement of data directly to or from application memory. This results in high bandwidth, low latency networking with little involvement from the CPU.

Today all the EDSFF family of form factors share the same protocol (NVMe), the same interface (PCIe), the same edge connector (SFF-TA-1002), and the same pinout and functions (SFF-TA-1009). There is a vast diversity of Enterprise and Datacenter applications. This presentation, from the 2020 OCP Summit, explains how having a flexible and scalable family of form factors allows for optimization for different use cases, different media types on SSDs, scalable performance, and improved data center TCO. the specification.

During the last two decades, the data center world has been moving to a “Software Defined Everything” paradigm. This has been taken care of mostly by hypervisors running on the x86 up to recently.

During the last two decades, the data center world has been moving to a “Software Defined Everything” paradigm. This has been taken care of mostly by hypervisors running on the x86 up to recently.

This presentation will discuss the work done to improve the timeout-recovery-abort protocol in NVMe. NVMexpress.org Technical Proposal 4097 made changes to the NVMe Abort Command and added a new IO Cancel Command to the NVMe 1.4  and 2.0 specifications. Support for these changes are being implemented in Data ONTAP, Linux, ESX, and SPDK.  This presentation will include a short demonstration of a working IO Cancel command implementation using a Linux host with an Data ONTAP NVMe/FC controller.