What Is Multiprotocol Label Switching?

Multi-Protocol Label Switching (English: Multi-Protocol Label Switching, abbreviated as MPLS ) is a new technology that uses labels to guide the high-speed and efficient transmission of data on an open communication network. The meaning of multi-protocol means that MPLS can not only support multiple protocols at the network layer level, but also be compatible with multiple data link layer technologies at the second layer.

Multi-Protocol Label Switching (MPLS) is a new-generation IP high-speed backbone network switching standard. It was proposed by the Internet Engineering Task Force (IETF).

MPLS uses labels for data forwarding. When a packet enters the network, it is necessary to assign a short tag of a fixed length to it and encapsulate the tag with the packet. During the entire forwarding process, the switching node forwards only based on the tag.

MPLS is independent of Layer 2 and Layer 3 protocols, such as ATM and IP. It provides a way to

MPLS: related signaling protocols, such as OSPF, BGP, ATM PNNI, etc.

LDP: Label Distribution Protocol

CR-LDP: based on

MPLS-based VPN

traditional

1. LDP and tradition

An MPLS network is an area composed of switching nodes running the MPLS protocol. These switching nodes are MPLS label switching routers. According to their positions in the MPLS network, they can be divided into MPLS label edge routers (LER: label edge router) and MPLS label core routers (LSR). As the name implies, LER is located at the edge of the MPLS network and is connected to other networks or users; LSR is located inside the MPLS network. The functions of the two types of routers differ slightly due to their location in the network.

In the MPLS architecture:

Control planes are based on connectionless services and are implemented using existing IP networks.

The forwarding plane (Forwarding Plane) is also called the data plane (Data Plane). It is connection-oriented and can use ATM,

LDP uses routing

MPLS, as a sorted forwarding technology, classifies packets with the same forwarding processing method into one class, called the Forwarding Equivalence Class (FEC). Packets of the same forwarding equivalence class will get exactly the same processing in the MPLS network.

A label is a short, fixed-length, local-only meaning

To learn more about the principles of MPLS, please refer to the following documents.

RFC3031: Multiprotocol Label Switching Architecture

1 Fully adopt the original IP routing and improve it on this basis; ensure the flexibility of MPLS network routing

2 Adopt ATM's efficient transmission and exchange mode, abandon the complex ATM signaling, and seamlessly integrate the advantages of IP technology into the efficient hardware forwarding of ATM

3 Data transmission and routing calculation are separated in MPLS network. It is a connection-oriented transmission technology that can provide effective QOS guarantee.

4 MPLS not only supports multiple network layer technologies, but also is a technology that has nothing to do with the link layer. It also supports X.25 Frame Relay ATM PPP SDH DWDM and so on. Different network transmission technologies are unified on the same MPLS platform

5 MPLS supports large-scale hierarchical network topology and has good network scalability

6 MPLS label merge mechanism supports combined transmission of different data streams

7 MPLS supports traffic engineering COS QOS and large-scale virtual private networks

MPLS is a label-based method of IP routing. These tags can be used to represent hop-by-hop or explicit routing, and indicate the quality of service (QoS), virtual private network, and the way in which a specific type of traffic (or a specific user's traffic) is transmitted on the network Class information. MPLS uses simplified technology to complete the conversion between Layer 3 and Layer 2. It can provide a label for each IP packet, and encapsulate it with the IP packet in a new MPLS packet, thereby determining the transmission path and priority of the IP packet. The MPLS-compatible router will Only the header tag of the MPLS data packet is read before forwarding according to the corresponding path, and there is no need to read the information such as the IP address bits in each IP data packet, so the data packet exchange and forwarding speed is greatly accelerated.

The current routing protocols choose the shortest path between a specified source and destination, regardless of the link status such as the bandwidth and load of the path, and there is no explicit method to bypass it for links that lack security. . With explicit routing, you have the flexibility to choose a low-latency, secure path to transmit data.

The MPLS protocol realizes the conversion from the routing of the third layer to the switching of the second layer. MPLS can use various Layer 2 protocols. The MPLS Working Group has so far standardized the tags used on Frame Relay, ATM and PPP links, and IEEE 802.3 LANs. One of the benefits of MPLS running on Frame Relay and ATM is that it is for these connection-oriented technologies.

Brings arbitrary connectivity to IP. At present, the main development direction of MPLS is in ATM. This is mainly because ATM has a strong traffic management function and can provide QoS services. The combination of ATM and MPLS technology can fully play its role in traffic management and QoS. The tag is the header used to forward the data packet. The format of the header depends on the network characteristics. In router networks, the tag is a separate 32-bit header; in ATM, the tag is placed in the virtual circuit identifier / virtual channel identifier (VCI / VPI) cell header. A key point for MPLS scalability is that the marking is only meaningful between the two devices communicating. At the core of the network, routers / switches only interpret tags and do not parse IP packets.

When an IP packet enters the core of the network, the border router assigns it a tag. From then on, the MPLS device will look at these tag information from beginning to end and exchange these tagged packets to their destination. As routing processing is reduced, network latency is reduced, while scalability is increased. The quality of service type of MPLS data packets can be determined by the MPLS border router according to various parameters of the IP packet, such as the source address, destination address, port number, and TOS value of the IP.

For IP packets arriving at the same destination, different forwarding paths can be established according to the requirements of their TOS values to meet their requirements for transmission quality. At the same time, through the management of special routes, it can also effectively solve the problem of load balancing and congestion in the network. When congestion occurs in the network, MPLS can establish new forwarding routes in real time to distribute traffic to alleviate network congestion.

MPLS switching uses a connection-oriented working mode. The connection-oriented working mode is that information transmission must go through the following three stages: connection establishment, data transmission, and disconnection. For MPLS, establishing a connection is the process of forming a label switched path LSP; data transmission is the process of forwarding data packets along the LSP; and tearing down the connection is the process of releasing the LSP when the communication ends or a fault occurs.

establish connection

(1) The way to drive connection establishment

MPLS technology supports three ways to drive the establishment of virtual connections: topology-driven, request-driven, and data-driven.

(2) Mark assignment

(3) Connection establishment process

(4) MPLS routing

data transmission

The data transmission of the MPLS network uses a label-based forwarding mechanism.

(1) Process of entry LER

When the data flow reaches the ingress LER, the ingress LER needs to complete three tasks: mapping the data packet to the LSP; encapsulating the data packet into a labeled packet; and forwarding the labeled packet from the corresponding port.

(2) Processing of LSR

LSR obtains the tag value from "SHIM", uses this tag value to index the LIB table, finds the output port and output tag of the corresponding entry, replaces the input tag with the output tag, and forwards it from the output port.

(3) Process of export LER

The egress router is the last node that the data packet goes through in the MPLS network, so the egress router needs to perform corresponding pop-up marking and other operations.

Remove the connection

Because the virtual connection in the MPLS network, that is, the LSP path is formed by connecting the bare metal channels identified by the tag in series, the removal of the connection is also the cancellation of the tag. There are two main ways to cancel the mark. One is to use a timer; the other is not to set a timer.

With the development of ASIC technology, routing lookup speed is no longer an obstacle

Cloud architecture

IaaS layer security mechanism through interface

MPLS uses traditional IP routing protocols based on distributed computing. When the network topology changes, the routes calculated by these protocols may cause loops instantly. Packets entering a looped LSP may cause two basic problems (1) the packet cannot be delivered to the correct destination address (2) congestion. After a loop occurs. Even if TTL subtraction and loop packet discarding are used, packets may still survive in the loop for a long time and occupy a lot of network resources. This has a significant impact on the correct transmission of other data packets that do not create a loop. The congestion caused by the loop data packets may cause the non-loop data packets to be lengthened or discarded, which may lead to network paralysis in severe cases.

In MPLS networks, there are many mechanisms to prevent loops. In terms of loop processing methods, the number of loops that may occur after using this method and the impact of this method on the convergence of routing calculations are generally considered. Reducing the occurrence of loops means that route convergence takes longer.

There are many methods for handling Layer 2 loops in MPLS networks, which can be divided into three categories:

1. Loop Survival This method minimizes the impact of loops on network service performance by, for example, limiting the size of the network resources that loops can use.

2. Loop detection allows loops to occur, but deletes them when loops are found in subsequent inspections.

3. Loop prevention to avoid loops in the Layer 2 forwarding path. ^[1]