MPLS

What is MPLS?

Multi-Protocol Label Switching (MPLS), proposed by the IETF, is a high-speed and highly efficient data communication technology that brings label exchange to the protocol. Multi-Protocol means MPLS supports multiple network layer protocols and complies with various data link layer technologies. Label Switching manifests: After data enters the MPLS domain, the label number will be pressed into the second and third layers so that when the routers in the domain forward the data packet, the label number based on the 2.5 layer only needs to query a local LFIB table (label forwarding information database). MPLS migrates the connectionless services to connection-oriented services. Label switching, control, and forwarding distributing are two critical technologies of MPLS.

How Does MPLS Work?

Control Panel

In the MPLS network, MPLS starts the Cisco private TDP or public LDP, and neighboring relationships between directly connected devices will be built.

LDP works based on UDP and CP646 ports. Sending multicast hello packet via UDP to discover neighbors. After acquiring the IP addresses of neighbors, TCP conversations will be built with the directly connected neighbors. The loopback address is often used to create TCP conversations and maintain reliable relationships. Then, make the route ID that the loopback address configures as MPLS. 

A neighbor table will be generated after the neighborhood is successfully established. LDP will generate labels based on routing lists learned from FIB.

The figures of labels are stored in local LIB tables (label information database). Since the LIB table will be shared among neighbors, the LIB table covers the label numbers distributed by the local and its neighbors for each routing.

Devices running MPLS protocols combine LIB with FIB and then generate LFIB tables for the relationship between label numbers and the optimal path (label forwarding information database).

Data Panel

When traffic enters RTB devices from the RTA devices, it queries the FIB table. The associated label number will directly confirm whether to press the labels. After reaching the RTC device, traffic will be forwarded based on the LFIB table if labels are pressed. Otherwise, it will be forwarded based on the FIB table. 

What are the Benefits of MPLS?

High-speed and highly efficient: MPLS saves the trouble of complex routing table queries by labeling packets for fast forwarding. Data transmission is faster and more efficient.

Agile Compatibility: MPLS can connect multiple types of networks and protocols, whether IPv4, IPv6, Ethernet, or ATM, and it can work seamlessly. Besides, it can help server vendors solve compatibility issues among different technologies.

High Reliability: MPLS has built-in redundancy (such as fast rerouting). When one link fails, the network service will switch to the redundant path, guaranteeing uninterrupted operation.

Better Performance and Low Latency: MPLS adapts to latency-critical applications such as audio, video, or mission-critical business data. The network can enable QoS by MPLS. By optimizing paths and setting priorities, MPLS provides better transmission performance.

Support traffic engineering: According to specific requirements, one can control the transmission patch, properly assign broadband, and avoid network congestion.

VPN and Scalable: MPLS supports setting up the virtual private network (VPN) and virtual leased line, making it easy for enterprises to scale networks. Organizations can also purchase broadband on demand and agilely adjust network resources.

What are the Cons of MPLS?

MPLS is costly compared to regular network services. LSP must be configured by MLSP vendors or organizations that apply MPLS, which is time-consuming to set up complex private paths in large-scale networks. These drawbacks prevent organizations from scaling networks in a limited time. MPLS doesn’t apply end-to-end encryption, which makes it vulnerable to network attacks. Any attacker that Intercepts packets in the MPLS path can read the packets in plain text. Besides, since organizations depending on cloud services are not accessible to specific servers where data and applications are stored, they may be unable to build direct network connections to their cloud servers.