Play with Tacker(by quqi99)

Install via devstack

#Enable tacker
BRANCH=master
enable_plugin heat https://git.openstack.org/openstack/heat $BRANCH
enable_plugin networking-sfc https://git.openstack.org/openstack/networking-sfc $BRANCH
enable_plugin barbican https://git.openstack.org/openstack/barbican $BRANCH
enable_plugin mistral https://git.openstack.org/openstack/mistral $BRANCH
enable_plugin tacker https://git.openstack.org/openstack/tacker $BRANCH

source openrc
export OS_USERNAME=admin
$ env |grep OS_
OS_PROJECT_DOMAIN_ID=default
OS_REGION_NAME=RegionOne
OS_USER_DOMAIN_ID=default
OS_PROJECT_NAME=demo
OS_IDENTITY_API_VERSION=3
OS_PASSWORD=password
OS_AUTH_TYPE=password
OS_AUTH_URL=http://172.16.1.1/identity
OS_USERNAME=admin
OS_TENANT_NAME=demo
OS_VOLUME_API_VERSION=2

Devstack会自动配置下列镜像与网络：

wget http://download.cirros-cloud.net/0.3.5/cirros-0.3.5-x86_64-disk.img openstack image create "cirros" --file cirros-0.3.5-x86_64-disk.img --disk-format qcow2 --container-format bare --public
$ glance image-list +--------------------------------------+--------------------------+
| ID | Name | +--------------------------------------+--------------------------+
| 55444fe2-4b26-4908-ba1b-4c5e95bd0b83 | cirros-0.3.5-x86_64-disk | | 26e711a9-d5c8-4991-a81d-5ebf5d2724d1 | OpenWRT | +--------------------------------------+--------------------------+ $ neutron net-list
neutron CLI is deprecated and will be removed in the future. Use openstack CLI instead. +--------------------------------------+----------+----------------------------------+-------------------------------------------------------+
| id | name | tenant_id | subnets | +--------------------------------------+----------+----------------------------------+-------------------------------------------------------+
| 43f2cbec-367c-45e8-bf37-26e66366b9eb | private  | 1e99096c578a4b2786c6a44ffd6bed44 | ab7d15c8-03d5-429a-926f-2ba07d0fb77f 10.0.0.0/26      |
| 57c739b9-6e61-4795-a72f-29d9f3ee5ab2 | net0     | b9c12c1af30349b987fe9ab638659e48 | ab177e6d-94ea-4d93-bbc6-d9e4c2bba90e 10.10.0.0/24     |
| 5c6dc85e-9fd0-49f6-8fc2-3386709b4a09 | net_mgmt | b9c12c1af30349b987fe9ab638659e48 | 7525b4e6-1bff-4b68-a2c6-d5633a155a7f 192.168.120.0/24 | | 63d88cd7-d540-446d-9538-4676b30cb494 | public | b9c12c1af30349b987fe9ab638659e48 | 7a2873fc-607b-4f9e-bd94-7ed91a746553 192.168.101.0/24 | | 95717e35-d3bc-4c6a-937c-b47c566a3666 | net1 | b9c12c1af30349b987fe9ab638659e48 | 0ecbc79b-c103-464f-ad59-2c7733c43e39 10.10.1.0/24 | +--------------------------------------+----------+----------------------------------+-------------------------------------------------------+

概念与步骤

Tacker中有这些概念：

VIM：相当于OpenStack
VNFD: VNF Descriptor
VNF: Virtual Network Function
所以依次需要：
创建VIM

cat > vim_config.yaml << EOF
auth_url: 'http://172.16.1.1/identity'
username: 'nfv_user'
password: 'devstack'
project_name: 'nfv'
project_domain_name: 'Default'
user_domain_name: 'Default'
EOF
tacker vim-register --is-default --config-file vim_config.yaml test-vim

创建vnfd

tacker vnfd-create --vnfd-file ./sample-vnfd.yaml sample-vnfd
tacker vnfd-list

创建vnf

tacker vnf-create --vnfd-name sample-vnfd sample-vnf

TOSCA VNFD语法

VNFD和juju的yaml类似,语法见下面sample-vnfd.yaml中的注释：

$ cat sample-vnfd.yaml 
tosca_deFinitions_version: tosca_simple_profile_for_nfv_1_0_0
description: Demo example
Metadata:
  template_name: sample-tosca-vnfd
topology_template:
  # VDU1 represents a node provided by underlying VIM
  node_templates:
    VDU1:
      type: tosca.nodes.nfv.VDU.Tacker
      # Define flavor
      capabilities:
        nfv_compute:
          properties:
            num_cpus: 1
            mem_size: 512 MB
            disk_size: 1 GB
      # Define image
      properties:
        image: cirros-0.3.5-x86_64-disk
        availability_zone: nova
        # Tacker has two integrated mgmt_driver to configure Metadata - noop and openwrt
        mgmt_driver: noop
        config: |
          param0: key1
          param1: key2
    # CP (Connection Point) represents a neutron port,CP1 is the name of this port
    CP1:
      type: tosca.nodes.nfv.CP.Tacker
      properties:
        management: true
        order: 0
        anti_spoofing_protection: false
      # One side of this CP is connecting to VL1,another side is connection to VDU1 
      requirements:
        - virtualLink:
            node: VL1
        - virtualBinding:
            node: VDU1
    # VL (Virtual Link) represents a neutron network
    VL1:
      type: tosca.nodes.nfv.VL
      properties:
        network_name: net_mgmt
        vendor: Tacker

VNFD的语法结构图如下：

Tacker如何实现水平伸缩

需有水平伸缩相关的配置：

# 定义策略，此节点与node_templates同级
  policies:
    # 此处定义了一个scale策略；SP1为自定义的名字
    - SP1:
        # scale策略的type
        type: tosca.policy.tacker.Scaling
        # 定义了策略的详细内容
        properties:
          # 每次伸缩的数量
          increment: 1
          # 下一次操作的冷却时间
          cooldown: 0
          # 总量的下限
          min_instances: 1
          # 总量的上限
          max_instances: 3
          # 初始的数量
          default_instances: 2
          # 要伸缩的目标，vdu、cp均可伸缩
          targets: [VDU1]

然后运行下列命令进行伸缩：

tacker vnf-scale --vnf-name test_vnf --scaling-policy-name SP1 --scaling-type out

上面伸缩配置和创建vim的配置一样会由Tacker最终转换为Heat模板，Heat模板最终再调用Neutron LBaaS v2来实现HA。点击这见我的另一博客。
而Juju的水平扩展则更为简单，直接运行如“juju add-unit nova-compute”之类的命令为自动增加节点，并且在这些节点之间添加haproxy+corosync的HA服务，点击这见我的另一博客。
如此说来，还是juju好用。

mgmt_driver

可以将一段脚本通过config参给VIM，然后tacker再ssh进这个VIM来执行这段配置。
这块juju则是直接ssh进VIM后自己该干嘛干嘛，灵活性更高。

Reference

[1] https://docs.openstack.org/tacker/latest/
[2] http://www.jb51.cc/article/p-obrdgoqa-bqg.html
[3] https://www.ibm.com/developerworks/cn/cloud/library/1511_zoupx_openstackheat/index.html
[4] https://www.ibm.com/developerworks/cn/cloud/library/cl-cloud-orchestration-technologies-trs/index.html