当前位置:网站首页>A new MPLS note from quigo, which must be read when taking the IE exam ---- quigo of Shangwen network
A new MPLS note from quigo, which must be read when taking the IE exam ---- quigo of Shangwen network
2022-07-28 23:01:00 【Shangwen network】
This technology sharing will help you understand MPLS VPN Related knowledge , As we all know, whether Huawei, Cisco or Ruijie MPLS VPN Is a required knowledge , Now let's understand MPLS VPN.

MPLS VPN Equipment name
CE User edge devices
PE Operator edge devices
P Backbone equipment of operators
characteristic
PE and PE Establish between VPN Tunnel ,CE Users don't need to know vpn Related content , Just a user
It's dynamic VPN,MPLS Of LSP It is a natural tunnel , And this tunnel is through LDP The protocol is dynamically established , So the difference is GRE VPN Something static VPN
Problems to be solved
Address space overlaps 
MPLS VPN working principle
Control plane ( Routing delivery )
CE1-PE1 The routing
Enable any routing protocol
PE How to deal with overlapping address spaces
VRF Virtual routing forwarding ( In essence, it is a physical PE The equipment becomes several dedicated PE, Each virtual private PE Connect to a user , fictitious PE Separate from each other )
Interface
agreement
Routing table
PE1-PE2 The routing
You have to use MPBGP
You can establish neighbors across routers
Support a large amount of routing information
High scalability , It can be easily compatible with new vpnv4 route
VPNv4 The birth of routing
VRF It is a concept of router locality , therefore PE1 Route to PE2 It's gone VRF, Cannot continue to distinguish conflicting routes
In order to distinguish conflicting routes in the delivery process , Added a new RD( Route identifier ), adopt RD The uniqueness of makes the route unique
RD Value guarantee is unique in the world
VPNv4=RD+IPv4
PE2-CE2 The routing
PE2 received PE1 After routing , How to determine which one you need to enter VRF example
adopt RT value
import RT
Compare the value when receiving the route
export RT
Carry this value when sending a route
Make sure to enter that VRF, Because of every VRF For one CE service , Then you can know the item is for that CE
Forwarding plane ( Message forwarding )
Outer label ( Public label )
Problem solved : When routing P The device does not participate in the transmission, and there is no relevant routing information , There will be a routing black hole problem
MPLS LDP The agreement came into being
PE2 The next hop of sending data is PE1, And because the LDP Produced PE1 and PE2 Between LSP, Therefore, tag forwarding is performed , here P The device will receive MPLS message , Perform tag forwarding
Inner label
Problem solved :PE2 Send the data to PE1 when ,PE1 How to know that you need to enter VRF Forward
Be careful :RT Cannot do this job ,RT Route attributes when routing for the control plane , Here is the data sending process , Package source IP、 Purpose IP、MAC Address 、 Port number, etc , non-existent RT
MPBGP Allocated
When PE1 adopt MPBGP Pass routing information to PE2 when , except RD、RT It will also carry a label .PE2 The tag can be carried when sending data , Opposite end PE1 Check the inner label after receiving the data , Naturally, you can know to give that VRF
The whole is here
Join a group to talk about technology , If you have any questions, you can share them directly QQ Group :316317765, You can also pay attention to the next official account. , More information will be released later .
----- Original by Kuige of Shangwen network 
边栏推荐
- 2020年国内十大IC设计企业曝光!这五大产业挑战仍有待突破!
- Leetcode exercise 3 - palindromes
- Console.log() console display... Solution
- 歌尔股份与上海泰矽微达成长期合作协议!专用SoC共促TWS耳机发展
- 《MySQL数据库进阶实战》读后感(SQL 小虚竹)
- No code development platform address book tutorial
- Pictures are named in batches in order (change size /jpg to PNG) [tips]
- 【三维目标检测】3DSSD(一)
- can‘t convert cuda:0 device type tensor to numpy. Use Tensor. cpu() to copy the tensor to host memory
- 《Shortening passengers’ travel time A dynamic metro train scheduling approach using deep reinforcem》
猜你喜欢

Yolov5 improvement 7: loss function improvement

What to do after mathematical modeling gets the competition problem and some ("crooked ways") tips - must see before the competition

console.log()控制台显示...解决办法

《Robust and Communication-Efficient Federated Learning From Non-i.i.d. Data》论文阅读

Multi activity disaster recovery construction after 713 failure of station B | takintalks share

Cglib create proxy

can‘t convert cuda:0 device type tensor to numpy. Use Tensor. cpu() to copy the tensor to host memory

Target detection notes fast r-cnn

【三维目标检测】3DSSD(二)

TypeError: can‘t convert cuda:0 device type tensor to numpy. Use Tensor. cpu() to copy the tensor to
随机推荐
Target detection notes - overview and common data sets
记录一下关于三角函数交换积分次序的一道题
Summary of C language learning content
Use FFT, matrix multiplication and conv2d to calculate convolution based on pytorch
Ocr-gan [anomaly detection: Reconstruction Based]
OSV_ q The size of tensor a (704) must match the size of tensor b (320) at non-singleton dime
Multi activity disaster recovery construction after 713 failure of station B | takintalks share
从 IPv4 向 IPv6 的迁移
Configuration and official document of Freia library [tips]
WebApplicationType#deduceFromClasspath
Es personal arrangement of relevant interview questions
One of the experience of solving problems with code error reporting: CUDA does not match pytorch
CGLIb 创建代理
希捷发布全新RISC-V架构处理器:机械硬盘相关性能暴涨3倍
Yolov5 improvement 15: network lightweight method deep separable convolution
Paper reading: deep forest / deep forest /gcforest
Introduction to address book export without code development platform
STM32 - Basic timer (tim6, tim7) working process, interpretation function block diagram, timing analysis, cycle calculation
Target detection notes -yolo
LTE小区搜索过程及SCH/BCH设计