Hcip day 14 (MPLS protocol)

Catalog

MPLS（ Multiprotocol label switching ）

Packet exchange

Label switching

Optimization process of packet switching

---

MPLS（ Multiprotocol label switching ）

Packet exchange

Data consists of packets , In each node of the network , The best way to reach the goal , That is, the process of routing and forwarding .

1、 Check two tables , That is, routing table and ARP Cache table .

2、 Matching principle of routing table （ The longest match principle ）.

3、 Recursive search .

4、IP Baotou variable length , That is, it can only be processed by software .

Control level ： The flow direction of data flow in the routing protocol , The purpose is to obtain the routing information of unknown network segments , Generate routing table .

Data level ： Based on the complete routing table of the equipment （FIB）, To forward specific data to the target network segment . The direction is just opposite to the control level .

Label switching

    Between two-layer package and three-layer package , Add a label that has a mapping relationship with the routing entry , Then maintain a record correspondence and forwarding interface table , Then carry the tag data to the device , You will see the label first , Then forward based on the maintained table , Second, no longer look at the third floor IP data . Because the label itself is short and fixed length , Therefore, the forwarding efficiency will be higher than that of packet switching . However, the label exchange process requires the action of pressing and ejecting labels , Therefore, the overall improvement of packet switching efficiency is not obvious .

Control level ： be based on FEC Assign tags , And get other LSR Yes, the same FEC Assigned tags . Record locally LIB In the table , combining FIB Table to generate LFIB surface .（ The functions of the control level can be completed through static manual configuration , It can also be done through dynamic protocol --- LDP agreement ： Label distribution agreement .）

Data level ： The device is based on LFIB surface , Forward according to the tag .

Optimization process of packet switching

1、 Process exchange ： The earliest packet switching , That is, every packet coming to the device needs to be based on the target IP Check the routing table and ARP Cache table for forwarding .

2、 Fast packet switching ： Stream based packet switching , One route, multiple exchanges --- Only the first packet needs to perform the packet exchange process .

3、 Cisco express exchange （Cisco Express Forwading）： namely CEF, For routing tables and ARP Pre read the contents of the cache table （ The lines in the routing table that need recursive search are recursive ） Recorded in the CEF In the table . And this CEF Tables can support hardware processing .

although CEF It's Cisco's proprietary technology , However, various manufacturers have developed their own express exchange technology based on the principle . When the guard equipment is transmitting data , You are not looking at the routing table , It is FIB surface （ Forwarding information database ）, This table supports hardware processing and forwarding .

     MPLS In fact, it is a label switching technology developed together with packet switching . Because tag generation needs to identify the three-tier protocol , and MPLS It's called multiprotocol label switching , Because it can recognize and compatible with a variety of three-tier protocols .

MPLS Main application areas ：

1、 Used to solve BGP The routing black hole of

2、MPLS-VPN

3、MPLS-TE（ Flow engineering ）, It can be simply understood as controlling the forwarding path of traffic .

    function MPLS A network of devices , We can call it a MPLS Domain . All running in the domain MPLS All devices can be called LSR（ Label switching router ）.

    This label is generated by each router based on the routing entries in its own routing table , It's the personal behavior of land carriers .

    stay MPLS in , We call packets with the same characteristics FEC（ Equivalent forwarding class ）, It is simply understood that all data packets matching the same route entry belong to the same FEC, One FEC Assign a label .

LIB surface （ Label information table ）

LFIB surface （ Tag information forwarding table ）, yes LIB Table and FIB The product of surface binding , The main record is the corresponding relationship between the tag number and the outgoing interface and the next hop .

    In the whole process , all MPLS In domain LSR Deal with yourself for a FEC In addition to the generated tags, you need to get other LSR Yes, the same FEC Assigned tags .

Combined with the above figure to analyze ：

R1 This device is for entering MPLS The data in the field is pressed into the label , So we call such a device ： Inbound LSR（Ingress LSR）

In the process ,R2 Completed a label replacement action , In this way LSR We call it ： transit LSR（transit LSR）

R4 yes MPLS A boundary of a field , What is completed is the pop-up action of the label , In this way LSR We call it ： To play LSR（egress LSR）

The path of data flow at the whole data level is R1-R2-R3-R4, We call this path LSP（ Label switching path ）

Generally one FEC It will correspond to a LSP. Particular attention ：LSP It's directional , If it is necessary to realize two-way data interworking , You have to build... In two directions LSP Talent .LSP How to build it ： static state 、 dynamic LDP

LABEL：20 position , The value range is 0-2 Of 20 Power . We call the value range of tags as tag space , The label space of each device is independent .

0-15, Special tag value , We usually don't use these labels when distributing , Because they have a special meaning

16-1023, Generally used for static LSP Build and use

1023-2 Of 20 Power ,LDP And the range of tag numbers used by protocols that can dynamically assign tags

EXP： Occupy 3 position , Mainly for strategy . In general 000, It can be understood as priority , The greater the numerical , The higher the priority , Can be forwarded first .

S： Occupy 1 position （ At the bottom of the stack ）, We call the ordered sequence of tag headers as tag stack . This bit is 1, It means the last label , by 0, It means that there are subsequent labels to be processed .

TTL： Is equivalent to TTL The value is counted in a different position , The function and purpose are the same .