update ovn eip nat doc (#110)

* update ovn eip nat doc * fix lint
kubeovn · Aug 3, 2023 · 51e5fd0 · 51e5fd0
1 parent 09d1d1d
commit 51e5fd0
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 30 deletions.
diff --git a/README.md b/README.md
@@ -3,4 +3,4 @@
 ```bash
 pip install -r docs/requirements.txt 
 mkdocs serve
-```
+```
diff --git a/docs/advance/ovn-eip-fip-snat.en.md b/docs/advance/ovn-eip-fip-snat.en.md
@@ -18,20 +18,20 @@ pod-->vpc1-subnet-->vpc1-->fip-->lrp-->external-subnet-->local-node-external-nic
 
 The pod access the public network based on the fip
 
-## 1. Deployment
+The CRD supported by this function is basically the same as the iptable nat gw public network solution.
 
-Currently allows all vpcs to share the same provider vlan subnet resources, similar to neutron ovn mode. Also compatible with previous scenarios **default vpc enable_eip_snat**.
+- ovn eip: occupies a public ip address and is allocated from the underlay provider network vlan subnet
+- ovn fip: one-to-one dnat snat, which provides direct public network access for ip addresses and vip in a vpc
+- ovn snat: a subnet cidr or a single vpc ip or vip can access public networks based on snat
+- ovn dnat: based router lb, which enables direct access to a group of endpoints in a vpc based on a public endpoint
 
-Executing install.sh requires specifying the default public logical switch.
+## 1. Deployment
 
-The design and use of this configuration item takes into account the following factors：
+Currently allows all vpcs to share the same provider vlan subnet resources, similar to neutron ovn mode.
+Compatible with previous scenarios [default VPC EIP/SNAT](../guide/eip-snat.en.md).
 
-- Based on this configuration item can be docked to the provider network, vlan, subnet resources.
-- Based on this configuration item, the default vpc enable_eip_snat function can be docked to the existing vlan, subnet resources, while supporting the ipam
-- If only the default vpc's enable_eip_snat mode is used with the old pod annotaion based eip fip snat, then the following configuration is not required.
-- Based on this configuration you can not use the default vpc enable_eip_snat process, only by corresponding to vlan, subnet process, can be compatible with only custom vpc use eip snat usage scenarios.
-
-The neutron ovn mode also has a certain static file configuration designation that is, for now, generally consistent.
+During the deployment phase, you may need to specify a default public network logical switch based on actual conditions.
+If no vlan is in use (vlan 0 is used), the following startup parameters do not need to be configured.
 
 ```bash
 # When deploying you need to refer to the above scenario and specify the following parameters as needed according to the actual situation
@@ -46,6 +46,15 @@ The neutron ovn mode also has a certain static file configuration designation th
 # currently only support to specify one underlay public network as the default external public network.
 ```
 
+The design and use of this configuration item takes into account the following factors：
+
+- Based on this configuration item can be docked to the provider network, vlan, subnet resources.
+- Based on this configuration item, the default vpc enable_eip_snat function can be docked to the existing vlan, subnet resources, while supporting the ipam
+- If only the default vpc's enable_eip_snat mode is used with the old pod annotaion based eip fip snat, then the following configuration is not required.
+- Based on this configuration you can not use the default vpc enable_eip_snat process, only by corresponding to vlan, subnet process, can be compatible with only custom vpc use eip snat usage scenarios.
+
+The neutron ovn mode also has a certain static file configuration designation that is, for now, generally consistent.
+
 ### 1.1 Create the underlay public network
 
 ``` bash
@@ -175,12 +184,11 @@ Route Table <main>:
 
 ## 2. ovn-eip
 
-This feature is basically the same as iptables-eip design and usage, ovn-eip currently has four types
+This function is designed and used in the same way as iptables-eip, ovn-eip currently has three types
 
-- lrp: Resources for vpc and public network connection
-- fip: For ovn nat dnat_and_snat resources
-- snat: For snat, supporting one-to-one to pod, and one-to-one throughout the subnet cidr
-- node-ext-gw: For ovn bfd-based ecmp routing scenarios
+- nat: indicates ovn dnat, fip, and snat. These nat types are recorded in status
+- lrp: indicates the resource used to connect a vpc to the public network
+- node-ext-gw: applies to ovn BFD-based ecmp static route scenario
 
 ``` bash
 ---

diff --git a/docs/advance/ovn-eip-fip-snat.md b/docs/advance/ovn-eip-fip-snat.md
@@ -1,4 +1,4 @@
-# OVN EIP FIP SNAT 支持
+# OVN EIP FIP SNAT DNAT 支持
 
 ``` mermaid
 graph LR
@@ -18,19 +18,21 @@ pod-->vpc1-subnet-->vpc1-->fip-->lrp-->external-subnet-->local-node-external-nic
 
 Pod 基于 FIP 出公网的大致流程，最后可以基于本地节点的公网网卡出公网。
 
-## 1. 部署
+该功能所支持的 CRD 在使用上将和 iptable nat gw 公网方案保持基本一致。
 
-目前允许所有（默认以及自定义）vpc 使用同一个 provider vlan subnet 资源，类似 neutron ovn 模式，同时兼容之前**默认 vpc 可以使用 enable_eip_snat**的场景。
+- ovn eip: 用于公网 ip 占位，从 underlay provider network vlan subnet 中分配
+- ovn fip： 一对一 dnat snat，为 vpc 内的 ip 或者 vip 提供公网直接访问能力
+- ovn snat：整个子网或者单个 vpc 内 ip 可以基于 snat 访问公网
+- ovn dnat：基于 router lb 实现, 基于公网 ip + 端口 直接访问 vpc 内的 一组 endpoints
 
-执行 install.sh 需要指定默认公网逻辑交换机。
-该配置项的设计和使用主要考虑了如下因素：
+## 1. 部署
 
-- 基于该配置项可以对接到 provider network，vlan，subnet 的资源。
-- 基于该配置项可以将默认 vpc enable_eip_snat 功能对接到已有的 vlan，subnet 资源，同时支持公网 ip 的 ipam。
-- 如果仅使用默认 vpc 的 enable_eip_snat 模式, 且仅使用旧的基于 pod annotaion 的 eip fip snat，那么这个配置无需配置。
-- 基于该配置可以不使用默认 vpc enable_eip_snat 流程，仅通过对应到 vlan，subnet 流程，可以兼容仅自定义 vpc 使用 eip snat 的使用场景。
+目前允许所有（默认以及自定义）vpc 使用同一个 provider vlan subnet 资源，同时兼容[默认 VPC EIP/SNAT](../guide/eip-snat.md)的场景。
+
+类似 neutron ovn，服务启动配置中需要指定 provider network 相关的配置，下述的启动参数也是为了兼容 VPC EIP/SNAT 的实现。
 
-neutron ovn 模式也有一定的静态文件配置指定设计，目前来说，大致一致。
+部署阶段，根据实际情况，可能需要指定默认公网逻辑交换机。
+如果实际使用中没有 vlan（使用 vlan 0），那么下述启动参数无需配置。
 
 ```bash
 # 部署的时候你需要参考以上场景，根据实际情况，按需指定如下参数
@@ -44,6 +46,13 @@ neutron ovn 模式也有一定的静态文件配置指定设计，目前来说
 ### 以上配置都和下面的公网网络配置 vlan id 和资源名保持一致，目前仅支持指定一个 underlay 公网作为默认外部公网。
 ```
 
+该配置项的设计和使用主要考虑了如下因素：
+
+- 基于该配置项可以对接到 provider network，vlan，subnet 的资源。
+- 基于该配置项可以将默认 vpc enable_eip_snat 功能对接到已有的 vlan，subnet 资源，同时支持公网 ip 的 ipam。
+- 如果仅使用默认 vpc 的 enable_eip_snat 模式, 且仅使用旧的基于 pod annotaion 的 eip fip snat，那么这个配置无需配置。
+- 基于该配置可以不使用默认 vpc enable_eip_snat 流程，仅通过对应到 vlan，subnet 流程，可以兼容仅自定义 vpc 使用 eip snat 的使用场景。
+
 ### 1.1 准备 underlay 公网网络
 
 ``` bash
@@ -173,12 +182,11 @@ Route Table <main>:
 
 ## 2. ovn-eip
 
-该功能和 iptables-eip 设计和使用方式基本一致，ovn-eip 目前有四种 type
+该功能和 iptables-eip 设计和使用方式基本一致，ovn-eip 目前有三种 type
 
+- nat: 用于 ovn dnat，fip, snat, 这些 nat 类型会记录在 status 中
 - lrp: 用于 vpc 和公网相连的资源
-- fip: 用于 ovn nat dnat_and_snat 资源
-- snat: 用于 snat，支持一对一到 pod ip，以及对应到 subnet cidr
-- node-ext-gw: 用于 ovn 基于 bfd 的 ecmp 路由场景
+- node-ext-gw: 用于 ovn 基于 bfd 的 ecmp 静态路由场景
 
 ``` bash
 ---