Pages

▼

Sunday, 27 June 2021

DPVS


DPVS is a high performance Layer-4 load balancer based on DPDK. It's derived from Linux Virtual Server LVS and its modification alibaba/LVS.

The name DPVS comes from "DPDK-LVS".

dpvs.png

Several techniques are applied for high performance:

  • Kernel by-pass (user space implementation).
  • Share-nothing, per-CPU for key data (lockless).
  • RX Steering and CPU affinity (avoid context switch).
  • Batching TX/RX.
  • Zero Copy (avoid packet copy and syscalls).
  • Polling instead of interrupt.
  • Lockless message for high performance IPC.
  • Other techs enhanced by DPDK.

Major features of DPVS including:

  • L4 Load Balancer, including FNAT, DR, Tunnel, DNAT modes, etc.
  • SNAT mode for Internet access from internal network.
  • NAT64 forwarding in FNAT mode for quick IPv6 adaptation without application changes.
  • Different schedule algorithms like RR, WLC, WRR, MH(Maglev Hashing), Conhash(Consistent Hashing) etc.
  • User-space Lite IP stack (IPv4/IPv6, Routing, ARP, Neighbor, ICMP ...).
  • Support KNI, VLAN, Bonding, Tunneling for different IDC environment.
  • Security aspect, support TCP syn-proxy, Conn-Limit, black-list， white-list.
  • QoS: Traffic Control.

DPVS feature modules are illustrated as following picture.

modules

Quick Start

Test Environment

This quick start is tested with the environment below.

  • Linux Distribution: CentOS 7.2
  • Kernel: 3.10.0-327.el7.x86_64
  • CPU: Intel(R) Xeon(R) CPU E5-2650 v3 @ 2.30GHz
  • NIC: Intel Corporation Ethernet Controller 10-Gigabit X540-AT2 (rev 03)
  • Memory: 64G with two NUMA node.
  • GCC: gcc version 4.8.5 20150623 (Red Hat 4.8.5-4)

Other environments should also be OK if DPDK works, please check dpdk.org for more info.

  • Please check this link for NICs supported by DPDK: http://dpdk.org/doc/nics.
  • Note flow-director (fdir) is needed for FNAT and SNAT mode with multi-cores.

Clone DPVS

$ git clone https://github.com/iqiyi/dpvs.git
$ cd dpvs

Well, let's start from DPDK then.

DPDK setup.

Currently, dpdk-stable-18.11.2 is recommended for DPVS. dpdk-stable-17.11.2 and dpdk-stable-17.11.6 are supported until the lifecycle end of DPVS v1.8.

You can skip this section if experienced with DPDK, and refer the link for details.

$ wget https://fast.dpdk.org/rel/dpdk-18.11.2.tar.xz   # download from dpdk.org if link failed.
$ tar xf dpdk-18.11.2.tar.xz

DPDK patchs

There are some patches for DPDK to support extra features needed by DPVS. Apply them if needed. For example, there's a patch for DPDK kni driver for hardware multicast, apply it if you are to launch ospfd on kni device.

Assuming we are in DPVS root directory and dpdk-stable-18.11.2 is under it, please note it's not mandatory, just for convenience.

$ cd <path-of-dpvs>
$ cp patch/dpdk-stable-18.11.2/*.patch dpdk-stable-18.11.2/
$ cd dpdk-stable-18.11.2/
$ patch -p1 < 0001-kni-use-netlink-event-for-multicast-driver-part.patch
$ patch -p1 < 0002-net-support-variable-IP-header-len-for-checksum-API.patch
$ ...

It's advised to patch all if your are not sure about what they are meant for.

DPDK build and install

Now build DPDK and export RTE_SDK env variable for DPDK app (DPVS).

$ cd dpdk-stable-18.11.2/
$ make config T=x86_64-native-linuxapp-gcc
Configuration done
$ make # or make -j40 to save time, where 40 is the cpu core number.
$ export RTE_SDK=$PWD
$ export RTE_TARGET=build

In our tutorial, RTE_TARGET is set to the default "build", thus DPDK libs and header files can be found in dpdk-stable-18.11.2/build.

Now to set up DPDK hugepage, our test environment is NUMA system. For single-node system please refer to the link.

$ # for NUMA machine
$ echo 8192 > /sys/devices/system/node/node0/hugepages/hugepages-2048kB/nr_hugepages
$ echo 8192 > /sys/devices/system/node/node1/hugepages/hugepages-2048kB/nr_hugepages

$ mkdir /mnt/huge
$ mount -t hugetlbfs nodev /mnt/huge

Install kernel modules and bind NIC with igb_uio driver. Quick start uses only one NIC, normally we use 2 for FNAT cluster, even 4 for bonding mode. For example, suppose the NIC we would use to run DPVS is eth0, in the meantime, we still keep another standalone NIC eth1 for debugging.

$ modprobe uio
$ cd dpdk-stable-18.11.2

$ insmod build/kmod/igb_uio.ko
$ insmod build/kmod/rte_kni.ko carrier=on

$ ./usertools/dpdk-devbind.py --status
$ ifconfig eth0 down  # assuming eth0 is 0000:06:00.0
$ ./usertools/dpdk-devbind.py -b igb_uio 0000:06:00.0

Note that a kernel parameter carrier is added to rte_kni.ko since DPDK v18.11, and the default value for it is "off". We need to load rte_kni.ko with the extra parameter carrier=on to make KNI devices work properly.

dpdk-devbind.py -u can be used to unbind driver and switch it back to Linux driver like ixgbe. You can also use lspci or ethtool -i eth0 to check the NIC PCI bus-id. Please refer to DPDK site for more details.

Note: PMD of Mellanox NIC is built on top of libibverbs using the Raw Ethernet Accelerated Verbs AP. It doesn't rely on UIO/VFIO driver. Thus, Mellanox NICs should not bind the igb_uio driver. Refer to Mellanox DPDK for details.

Build DPVS

It's simple, just set RTE_SDK and build it.

$ cd dpdk-stable-18.11.2/
$ export RTE_SDK=$PWD
$ cd <path-of-dpvs>

$ make # or "make -j40" to speed up.
$ make install

Build dependencies may be needed, such as automake, libnl3, libnl-genl-3.0, openssl, popt and numactl. You can install the missing dependencies by using the package manager of the system, e.g., yum install popt-devel (CentOS).

Output files are installed to dpvs/bin.

$ ls bin/
dpip  dpvs  ipvsadm  keepalived
  • dpvs is the main program.
  • dpip is the tool to set IP address, route, vlan, neigh, etc.
  • ipvsadm and keepalived come from LVS, both are modified.

Launch DPVS

Now, dpvs.conf must locate at /etc/dpvs.conf, just copy it from conf/dpvs.conf.single-nic.sample.

$ cp conf/dpvs.conf.single-nic.sample /etc/dpvs.conf

and start DPVS,

$ cd <path-of-dpvs>/bin
$ ./dpvs &

Check if it's get started ?

$ ./dpip link show
1: dpdk0: socket 0 mtu 1500 rx-queue 8 tx-queue 8
    UP 10000 Mbps full-duplex fixed-nego promisc-off
    addr A0:36:9F:9D:61:F4 OF_RX_IP_CSUM OF_TX_IP_CSUM OF_TX_TCP_CSUM OF_TX_UDP_CSUM

If you see this message. Well done, DPVS is working with NIC dpdk0!

Don't worry if you see this error:

EAL: Error - exiting with code: 1
  Cause: ports in DPDK RTE (2) != ports in dpvs.conf(1)

It means the NIC count of DPVS does not match /etc/dpvs.conf. Please use dpdk-devbind to adjust the NIC number or modify dpvs.conf. We'll improve this part to make DPVS more "clever" to avoid modify config file when NIC count does not match.

What config items does dpvs.conf support and how to configure them? Well, DPVS maintains a config item file conf/dpvs.conf.items which lists all supported config entries and corresponding feasible values.

Test Full-NAT (FNAT) Load Balancer

The test topology looks like the following diagram.

fnat-single-nic

Set VIP and Local IP (LIP, needed by FNAT mode) on DPVS. Let's put commands into setup.sh. You do some check by ./ipvsadm -ln, ./dpip addr show.

$ cat setup.sh
VIP=192.168.100.100
LIP=192.168.100.200
RS=192.168.100.2

./dpip addr add ${VIP}/24 dev dpdk0
./ipvsadm -A -t ${VIP}:80 -s rr
./ipvsadm -a -t ${VIP}:80 -r ${RS} -b

./ipvsadm --add-laddr -z ${LIP} -t ${VIP}:80 -F dpdk0
$
$ ./setup.sh

Access VIP from Client, it looks good!

client $ curl 192.168.100.100
Your ip:port : 192.168.100.3:56890

Tutorial Docs

More configure examples can be found in the Tutorial Document. Including,

  • WAN-to-LAN FNAT reverse proxy.
  • Direct Route (DR) mode setup.
  • Master/Backup model (keepalived) setup.
  • OSPF/ECMP cluster model setup.
  • SNAT mode for Internet access from internal network.
  • Virtual Devices (Bonding, VLAN, kni, ipip/GRE).
  • UOA module to get real UDP client IP/port in FNAT.
  • ... and more ...

We also listed some frequently asked questions in the FAQ Document. It may help when you run into problems with DPVS.

from https://github.com/iqiyi/dpvs

------

相关帖子：https://briteming.blogspot.com/2021/01/my-dpdkdns.html

at

No comments:

Post a Comment