NVMe over Fabrics with SPDK with iRDMA总结 - 1

1.0     Introduction简介

NVM Express* (NVMe*) drives are high-speed, low-latency, solid-state drives (SSDs), that connect over the server Peripheral Component Interconnect Express* (PCIe*) bus. NVM Express* (NVMe*) 硬盘是高速、低延迟的固态硬盘 (SSD),通过服务器外设组件互连 Express* (PCIe*) 总线连接。

The development of these high-performance drives has spurred new innovation in storage over networking protocols, which takes full advantage of the drive capabilities in data center and cloud environments. 这些高性能硬盘的开发推动了网络存储协议的创新,从而充分利用了硬盘在数据中心和云环境中的功能。

NVMe over Fabrics (NVMe-oF) provides networked storage at a latency level close to locally-mounted storage through a re-architected storage protocol that combines the use of low-latency/high-efficiency fabric technologies such as Remote Direct Memory Access (RDMA) or Fibre Channel (FC) with these high-speed NVMe drives. NVMe over Fabrics(NVMe-oF)通过重新架构的存储协议,将低延迟/高效率的技术(如远程直接内存访问(RDMA)或光纤通道(FC))与这些高速 NVMe 硬盘结合使用,以接近本地安装存储的延迟水平提供网络存储。

The Storage Performance Development Kit (SPDK) provides a set of tools and libraries for writing high-performance, scalable, user-mode storage applications. SPDK is an open-source project focused on optimizing storage software for the latest generation CPUs, NVMe SSDs, and NICs to improve the performance and efficiency of storage applications. It achieves high-performance by moving all of the necessary storage drivers into userspace and operating in a polled mode instead of interrupts, which avoids kernel context switches and eliminates interrupt handling overhead. 存储性能开发工具包(SPDK)为编写高性能、可扩展的用户模式存储应用提供了一套工具和库。SPDK 是一个开源项目,专注于为最新一代 CPU、NVMe SSD 和网卡优化存储软件,以提高存储应用的性能和效率。它将所有必要的存储驱动程序移至用户空间,并以轮询模式而非中断模式运行,从而避免了内核上下文切换并消除了中断处理开销,从而实现了高性能。

The SPDK community started with a userspace, polled-mode, asynchronous, lock-less NVMe driver and then extended the performance and efficiencies of SPDK to storage networking and virtualization. SPDK provides a high-performant NVMe-oF target and host components that are spec compliant. Therefore, users can use the Linux Kernel NVMe-oF host to connect to an SPDK NVMe-oF target and vice versa. SPDK 社区从用户空间、轮询模式、异步、无锁 NVMe 驱动程序起步,然后将 SPDK 的性能和效率扩展到存储网络和虚拟化。SPDK 提供了符合规范的高性能 NVMe-oF 目标机和主机组件。因此,用户可以使用 Linux 内核 NVMe-oF 主机连接 SPDK NVMe-oF 目标机,反之亦然。

Intel supports NVMe over Fabrics on two Intel® Ethernet product lines with RDMA technology: 英特尔在两个采用 RDMA 技术的英特尔®以太网产品线上支持 NVMe over Fabrics:

  • Intel® Ethernet 800 Series
  • Intel® Ethernet Connection X722/Intel® Ethernet Network Adapter X722.

1.1       Purpose目的

This document is a reference guide for configuring the SPDK NVMe over Fabrics target application on the Linux* operating systems using Intel® Ethernet 800 Series or Intel® Ethernet Connection X722/ Intel® Ethernet Network Adapter X722. 本文档是在 Linux* 操作系统上使用英特尔®以太网 800 系列或英特尔®以太网连接 X722/ 英特尔®以太网网络适配器 X722 配置 SPDK NVMe over Fabrics 目标应用程序的参考指南。

相关推荐

  1. 1.27学习总结

    2023-12-14 01:54:02       45 阅读
  2. 1.30学习总结

    2023-12-14 01:54:02       45 阅读
  3. 学习总结——1.31

    2023-12-14 01:54:02       67 阅读
  4. Pytorch 复习总结 1

    2023-12-14 01:54:02       42 阅读

最近更新

  1. docker php8.1+nginx base 镜像 dockerfile 配置

    2023-12-14 01:54:02       94 阅读
  2. Could not load dynamic library ‘cudart64_100.dll‘

    2023-12-14 01:54:02       100 阅读
  3. 在Django里面运行非项目文件

    2023-12-14 01:54:02       82 阅读
  4. Python语言-面向对象

    2023-12-14 01:54:02       91 阅读

热门阅读

  1. 区块链是个啥

    2023-12-14 01:54:02       52 阅读
  2. 小tips: add pub key to GitHub

    2023-12-14 01:54:02       59 阅读
  3. 数据结构和算法专题---3、失效算法与应用

    2023-12-14 01:54:02       41 阅读
  4. 常见的响应状态码

    2023-12-14 01:54:02       63 阅读
  5. C++(11):判断tuple是否含有某个类型

    2023-12-14 01:54:02       63 阅读
  6. bash命令: logrotate的使用

    2023-12-14 01:54:02       56 阅读
  7. 06组团队项目-Beta冲刺-1/3

    2023-12-14 01:54:02       59 阅读
  8. 前端成神之路-CSS基础选择器

    2023-12-14 01:54:02       57 阅读
  9. Playwright:下一代自动化测试工具

    2023-12-14 01:54:02       51 阅读
  10. 五花八门客户问题(BUG) - 用好strace2

    2023-12-14 01:54:02       58 阅读
  11. Redis常问面试题

    2023-12-14 01:54:02       71 阅读