Local模式：一般测试时使用，只需一台物理机即可。
GRE模式：隧道数量没有限制，性能有点问题。
Vlan模式：vlan数量有4096的限制
VXlan模式：vlan数量没有限制，性能比GRE好。
Flat模式：管理员创建租户直接到外网，不需要NAT了。
目前的几个项目正好用到了这种Flat模式的网络结构，我觉得小项目中还是比较方便的，Openstack的体系和框架越来越稳定，项目中用用虚拟化和存储，直接借助Openstack还是很方便。这时网络结构可能就不需要用租户、或者在细分其它，Flat模式反而更简洁直观，毕竟外网IP直接到了instance。
下面以我实验环境配置为例，存储部分配置省略。
环境：controller node一台       compute node一台       OS：Centos6.4
系统逻辑结构如下

一、controller node 配置
1、网络
[root@controller ~]# more /etc/hosts
192.168.3.98    controller
192.168.3.95    compute
[root@controller ~]# more /etc/sysconfig/network-scripts/ifcfg-eth0   #外部网络
DEVICE=eth0
HWADDR=10:BF:48:B7:8F:A0
TYPE=Ethernet
ONBOOT=yes
BOOTPROTO=none
IPADDR=192.168.3.98
NETMASK=255.255.255.0
GATEWAY=192.168.3.1
[root@controller ~]# more /etc/sysconfig/network-scripts/ifcfg-eth1  #vm 及node之间通讯使用
DEVICE=eth1
TYPE=Ethernet
ONBOOT=yes
[root@controller ~]#  more /etc/sysconfig/network
NETWORKING=yes
HOSTNAME=controller
GATEWAY=192.168.3.1
[root@controller ~]#  more /etc/resolv.conf
nameserver 202.102.134.68
nameserver 202.102.128.68

2、添加源
yum install wget
cd /etc/yum.repos.d/
wget
yum list

3、LVM配置 [我单独拿出一个分区给cinder使用，同时我的控制和计算节点也都是cinder节点]
[root@controller ~]# umount /dev/sda5
[root@controller ~]# pvcreate /dev/sda5Physical volume "/dev/sda5" successfully created
[root@controller ~]# vgcreate cinder-volumes /dev/sda5Volume group "cinder-volumes" successfully created

4、安装packstack
yum install -y
yum install -y openstack-packstack
yum -y update
重启机器

5、Packstack配置
创建一个应答文件

packstack --gen-answer-file=multi-node.txt

修改应答文件，由于我已经有专门的cinder volume, 所以不需要创建。如果没有手工创建cinder volume，那么这就不修改就可以。

CONFIG_CINDER_VOLUMES_CREATE=n              #不在创建volume
CONFIG_HEAT_INSTALL=y
CONFIG_NAGIOS_INSTALL=y
CONFIG_KEYSTONE_ADMIN_PW=admin
CONFIG_NOVA_COMPUTE_HOSTS=192.168.3.98  #此处可以添加compute node IP ,号分割
CONFIG_NEUTRON_OVS_TENANT_NETWORK_TYPE=vlan
CONFIG_NEUTRON_OVS_VLAN_RANGES=physnet1:100:200
CONFIG_NEUTRON_OVS_BRIDGE_MAPPINGS=physnet1:br-eth1
CONFIG_NEUTRON_OVS_BRIDGE_IFACES=br-eth1:eth1
CONFIG_HEAT_CLOUDWATCH_INSTALL=y
CONFIG_HEAT_CFN_INSTALL=y
CONFIG_NAGIOS_PW=admin

6、安装 [RDO的方式部署简直太简单了，我值截取部分]
[root@controller ~]# packstack --answer-file=multi-node.txt           #安装过程，开始会输入密码，多节点的话会依此类推。安装过程有可能因为源的原因中断，可以重复运行。
Welcome to Installer setup utility
Packstack changed given value  to required value /root/.ssh/id_rsa.pub

Installing:
Clean Up...                                            [ DONE ]

7、 装完后，有可能由于RDO的问题，没有创建br-ex，需要我们手工来创建。
【目前看2014年后，redhat已经修复这个bug，剩下的只是我们对应的创建网卡配置文件和启动port】
[root@controller network-scripts]# more ifcfg-br-ex
DEVICE=br-ex
IPADDR=192.168.3.98
PREFIX=24
GATEWAY=192.168.3.1
DNS1=202.102.134.68
ONBOOT=yes
修改 /etc/sysconfig/network-scripts/ifcfg-eth0,  注意：一定要加上eth0 的mac地址，不然会出错。

[root@controller network-scripts]# more ifcfg-eth0
DEVICE=eth0
HWADDR=00:90:27:E0:61:14
#TYPE=Ethernet
ONBOOT=yes
#BOOTPROTO=none
#IPADDR=192.168.3.98
#NETMASK=255.255.255.0

ovs-vsctl add-port br-ex eth0; service network restart
当运行 ovs-vsctl add-port的时候，网络就会中断，所以你两条命令需要一起执行。

8、改变桥接网络配置，此处主要修改neutron配置，我们的flat网络模式是依赖在vlan的基础上改进的，也是此文重点之处。
编辑
vi /etc/neutron/plugins/openvswitch/ovs_neutron_plugin.ini
原安装完初始内容如下
[OVS]
network_vlan_ranges=physnet1:100:200
tenant_network_type=vlan
enable_tunneling=False
integration_bridge=br-int
bridge_mappings=physnet1:br-eth1

修改成

[OVS]
network_vlan_ranges=physnet1,physnet2:100:200
tenant_network_type=vlan
enable_tunneling=False
integration_bridge=br-int
bridge_mappings=physnet1:br-ex,physnet2:br-eth1

然后重启相关服务，如果为省事，重启机器
之后检查网络和桥接情况查询
[root@controller ~]# ovs-vsctl show

9、创建Flat模式网络，目前都可以在web界面操作
[root@controller ~(keystone_admin)]# neutron net-create Public --provider:network_type flat --provider:physical_network physnet1
Created a new network:
+---------------------------+--------------------------------------+
| Field                     | Value                                |
+---------------------------+--------------------------------------+
| admin_state_up            | True                                 |
| id                        | 3c5eca6f-aeb4-451b-9fa2-ed70d8bab9e9 |
| name                      | Public                               |
| provider:network_type     | flat                                 |
| provider:physical_network | physnet1                             |
| provider:segmentation_id  |                                      |
| shared                    | False                                |
| status                    | ACTIVE                               |
| subnets                   |                                      |
| tenant_id                 | 2cfb66b8a9524cc68c2e1bf424bda943     |
+---------------------------+--------------------------------------+
[root@controller ~(keystone_admin)]# neutron subnet-create Public 192.168.3.0/24 --disable-dhcp --allocation-pool start=192.168.3.100,end=192.168.3.110
Created a new subnet:
+------------------+----------------------------------------------------+
| Field            | Value                                              |
+------------------+----------------------------------------------------+
| allocation_pools | {"start": "192.168.3.100", "end": "192.168.3.110"} |
| cidr             | 192.168.3.0/24                                     |
| dns_nameservers  |                                                    |
| enable_dhcp      | False                                              |
| gateway_ip       | 192.168.3.1                                        |
| host_routes      |                                                    |
| id               | 8d089754-3764-4249-98f2-402f62a872bc               |
| ip_version       | 4                                                  |
| name             |                                                    |
| network_id       | 3c5eca6f-aeb4-451b-9fa2-ed70d8bab9e9               |
| tenant_id        | 2cfb66b8a9524cc68c2e1bf424bda943                   |
+------------------+----------------------------------------------------+

10、cinder配置
查询部署的服务是否正常
[root@controller ~(keystone_admin)]# cinder-manage service list
Binary           Host                                 Zone             Status     State Updated At
cinder-volume    controller                           nova             enabled        2014-02-10 07:42:22
cinder-scheduler controller                           nova             enabled        2014-02-10 07:42:30
[root@controller ~(keystone_admin)]# cinder-manage host list
host                            zone
controller                      nova

然后查询论证是否成功
[root@controller ~(keystone_admin)]# lvdisplay --- Logical volume ---LV Path                /dev/cinder-volumes/volume-0a9cc8fc-d3fd-4dd1-9501-9c3813056415LV Name                volume-0a9cc8fc-d3fd-4dd1-9501-9c3813056415VG Name                cinder-volumesLV UUID                zbGuJa-3bdk-V6UZ-ZBsx-FWeA-dlys-qMphasLV Write Access        read/writeLV Creation host, time controller, 2014-02-10 15:46:44 +0800LV Status              available# open                 1LV Size                5.00 GiBCurrent LE             1280Segments               1Allocation             inheritRead ahead sectors     auto- currently set to     4096Block device           253:0

[root@controller ~(keystone_admin)]# cinder list
+--------------------------------------+-----------+--------------+------+-------------+----------+-------------+
|                  ID                  |   Status  | Display Name | Size | Volume Type | Bootable | Attached to |
+--------------------------------------+-----------+--------------+------+-------------+----------+-------------+
| 0a9cc8fc-d3fd-4dd1-9501-9c3813056415 | available |     oak      |  5   |     SSD     |  false   |             |
+--------------------------------------+-----------+--------------+------+-------------+----------+-------------+
[root@controller ~(keystone_admin)]#

此时，启动一个instance
可以在云硬盘中，选择挂载，挂载到相应的instance上。

二、compute node 配置
1、网络，参考控制节点，其次检查ntp，时间是否与控制节点同步
2、添加源和Openstack软件环境都参考控制节点的前期准备
3、RDO应答文件，填写了compute node IP后，一般nova-compute、openstack-neutron-openvswitch和MySQL-python都已经安装
4、因为我的环境是Flat模式，所以此处neutron配置文件还需修改，然后创建相应的桥
修改neutron
vi /etc/neutron/plugins/openvswitch/ovs_neutron_plugin.ini
[OVS]
network_vlan_ranges=physnet1,physnet2:100:200
tenant_network_type=vlan
enable_tunneling=False
integration_bridge=br-int
bridge_mappings=physnet1:br-ex,physnet2:br-eth1

创建桥
[root@compute neutron]# cd /etc/sysconfig/network-scripts/
[root@compute network-scripts]# more ifcfg-br-ex
DEVICE=br-ex
IPADDR=192.168.3.95
PREFIX=24
GATEWAY=192.168.3.1
DNS1=202.102.134.68
ONBOOT=yes
[root@compute network-scripts]# more ifcfg-eth0
DEVICE=eth0
HWADDR=00:90:27:E0:61:14
#TYPE=Ethernet
ONBOOT=yes
#BOOTPROTO=none
#IPADDR=192.168.3.95
#NETMASK=255.255.255.0
#GATEWAY=192.168.3.1
[root@compute network-scripts]# ovs-vsctl add-br br-ex
[root@compute network-scripts]# ovs-vsctl add-port br-ex eth0; service network restart

[root@compute network-scripts]# ovs-vsctl show
3875869a-c249-4514-aadc-b95b92e7f304Bridge br-exPort br-exInterface br-extype: internalPort "eth0"Interface "eth0"Bridge br-intPort br-intInterface br-inttype: internalovs_version: "1.11.0"

5、cinder 配置
[root@compute ~]# umount /dev/sda4
[root@compute ~]# pvcreate /dev/sda4Physical volume "/dev/sda4" successfully created
[root@compute ~]# vgcreate cinder-volumes /dev/sda4Volume group "cinder-volumes" successfully created

yum install openstack-cinder openstack-utils

vi  /etc/cinder/api-paste.ini 参考拷贝控制节点内容
[filter:authtoken]
paste.filter_factory = keystoneclient.middleware.auth_token:filter_factory
# signing_dir is configurable, but the default behavior of the authtoken
# middleware should be sufficient.  It will create a temporary directory
# in the home directory for the user the cinder process is running as.
#signing_dir = /var/lib/cinder/keystone-signing

admin_tenant_name=services
auth_host=192.168.3.98
service_port=5000
auth_uri=http://192.168.3.98:5000/
auth_port=35357
service_host=192.168.3.98
service_protocol=http
admin_user=cinder
auth_protocol=http
admin_password=570fe4a9f7bd4e2d

# openstack-config --set /etc/cinder/cinder.conf DEFAULT rpc_backend cinder.openstack.common.rpc.impl_qpid
# openstack-config --set /etc/cinder/cinder.conf DEFAULT qpid_hostname 192.168.3.98

vi  /etc/cinder/api-paste.ini 参考拷贝控制节点内容

sql_connection=mysql://cinder:0ba1a335f2834e1a@192.168.3.98/cinder
qpid_reconnect_timeout=0
qpid_reconnect_limit=0
qpid_reconnect=True
qpid_reconnect_interval_max=0
qpid_reconnect_interval_min=0
sql_idle_timeout=3600
qpid_reconnect_interval=0
notification_driver=cinder.openstack.common.notifier.rpc_notifier

vi /etc/tgt/targets.conf  增加如下内容

include /etc/cinder/volumes/*

# service openstack-cinder-volume start
# service tgtd start
# chkconfig openstack-cinder-volume on
# chkconfig tgtd on

[root@controller ~(keystone_admin)]# cinder-manage service list
Binary           Host                                 Zone             Status     State Updated At
cinder-volume    controller                           nova             enabled        2014-02-11 06:31:08
cinder-scheduler controller                           nova             enabled        2014-02-11 06:31:06
cinder-volume    compute                              nova             enabled        2014-02-11 06:30:40
[root@controller ~(keystone_admin)]#