Storage Performance Development Kit - http://spdk.io
C Makefile Shell Python Other
Latest commit 5ee4728 Jan 23, 2017 @uchida2016 uchida2016 committed with dverkamp-intel log: define prioritynames[] (#102)
Permalink
Failed to load latest commit information.
app nvmf_tgt: check existence of bdev in construct_nvmf_subsystem RPC Jan 18, 2017
build/lib build: consolidate library outputs in build/lib Nov 17, 2016
doc doc: point Documentation link at index page Jan 18, 2017
etc/spdk bdev/nvme: add timer-based admin queue poller Jan 19, 2017
examples nvme/perf: allow multiple -r (transport ID) arguments Jan 21, 2017
include nvme: mention TRID parsing function in struct docs Jan 20, 2017
lib log: define prioritynames[] (#102) Jan 23, 2017
mk build: enable -fno-common Jan 20, 2017
scripts scripts/genconfig.py: fix format specs for older Python Jan 13, 2017
test log: define prioritynames[] (#102) Jan 23, 2017
.astylerc build: check formatting with astyle Sep 23, 2015
.gitignore gitignore: ignore .kdev4 (KDevelop) files Jul 12, 2016
.travis.yml readme, travis: update to DPDK 16.11 Nov 15, 2016
CHANGELOG.md doc: move porting guide into doc/ and add to index Jan 17, 2017
CONFIG config: add option to turn on UBSan Dec 6, 2016
LICENSE Remove year from copyright headers. Jan 28, 2016
Makefile build: generate config.h and implicitly include it Jun 8, 2016
README.md doc: move porting guide into doc/ and add to index Jan 17, 2017
autobuild.sh autobuild: log current git version and date Dec 16, 2016
autopackage.sh autopackage: enable -Werror in release build Nov 17, 2016
autorun.sh eofnl: check for extra trailing newlines Oct 11, 2016
autotest.sh autotest: stop excluding test/ from lcov results Jan 21, 2017
unittest.sh test/nvme: run quirks unit test Jan 21, 2017

README.md

Storage Performance Development Kit

Build Status

SPDK Mailing List

SPDK on 01.org

The Storage Performance Development Kit (SPDK) provides a set of tools and libraries for writing high performance, scalable, user-mode storage applications. It achieves high performance by moving all of the necessary drivers into userspace and operating in a polled mode instead of relying on interrupts, which avoids kernel context switches and eliminates interrupt handling overhead.

The development kit currently includes:

Documentation

Doxygen API documentation is available, as well as a Porting Guide for porting SPDK to different frameworks and operating systems.

Many examples are available in the examples directory.

Changelog

Prerequisites

To build SPDK, some dependencies must be installed.

Fedora/CentOS:

sudo dnf install -y gcc gcc-c++ CUnit-devel libaio-devel openssl-devel
# Additional dependencies for NVMe over Fabrics:
sudo dnf install -y libibverbs-devel librdmacm-devel

Ubuntu/Debian:

sudo apt-get install -y gcc g++ make libcunit1-dev libaio-dev libssl-dev
# Additional dependencies for NVMe over Fabrics:
sudo apt-get install -y libibverbs-dev librdmacm-dev

FreeBSD:

  • gcc
  • gmake
  • cunit
  • openssl

Additionally, DPDK is required.

1) cd /path/to/spdk
2) wget http://fast.dpdk.org/rel/dpdk-16.11.tar.xz
3) tar xf dpdk-16.11.tar.xz

Linux:

4) (cd dpdk-16.11 && make install T=x86_64-native-linuxapp-gcc DESTDIR=.)

FreeBSD:

4) (cd dpdk-16.11 && gmake install T=x86_64-native-bsdapp-clang DESTDIR=.)

Building

Once the prerequisites are installed, run 'make' within the SPDK directory to build the SPDK libraries and examples.

make DPDK_DIR=/path/to/dpdk

If you followed the instructions above for building DPDK:

Linux:

make DPDK_DIR=./dpdk-16.11/x86_64-native-linuxapp-gcc

FreeBSD:

gmake DPDK_DIR=./dpdk-16.11/x86_64-native-bsdapp-clang

Hugepages and Device Binding

Before running an SPDK application, some hugepages must be allocated and any NVMe and I/OAT devices must be unbound from the native kernel drivers. SPDK includes a script to automate this process on both Linux and FreeBSD. This script should be run as root.

sudo scripts/setup.sh

Examples

Example code is located in the examples directory. The examples are compiled automatically as part of the build process. Simply call any of the examples with no arguments to see the help output. You'll likely need to run the examples as a privileged user (root) unless you've done additional configuration to grant your user permission to allocate huge pages and map devices through vfio.