Key Technologies of PTN - Packet Forwarding Technology

1. Key Technologies of PTN - Packet Forwarding Technology V1.0

2. Contents • Evolution History of PTN Technology • MPLS Technology • MPLS-TP Technology

3. MPLS/ Enhanced Ethernet Evolution History of PTN Technology Early Stage Present Situation MPLS-TP SDH Like OAM/PS • During the early stage of development, there were two types of PTN technologies: T-MPLS and PBT. • T-MPLS: The T-MPLS technology is evolved from the MPLS technology, and it complies with the features of transmission. • PBT: The PBT technology is evolved from the Ethernet technology. This connection-oriented packet transmission technology adopts the MAC-in-MAC technology, it closes the MAC address self-learning function for the operator and adds the configurations of network administration management and network control. It was introduced by Nortel, and it was supported by very few manufacturers. Nowadays, this technology seems to have no users any more. • Along with the technology development, the T-MPLS technology has evolved to be the MPLS-TP technology with the improvement on OAM. • Nowadays, the MPLS-TP technology is widely used as a key technology of PTN.

4. Contents • Evolution History of PTN Technology • MPLS Technology • MPLS-TP Technology

5. Features of Traditional IP Technology: The IP communication is conducted in the hop-by-hop manner; The rule of "longest matching" is applied in packet forwarding; The network devices need to know the routes of the entire network, otherwise they cannot forward the packets in the network segment; QoS cannot be guaranteed: Since IP protocol is a non-connection protocol, the concept of QoS is not applicable to the Internet. The throughput and transmission delay cannot be guaranteed, the network can only take the best effort to satisfy the users' needs. New services cannot be implemented in large scale unless the network environment is improved. Background of MPLS Technology

6. Features of ATM Forwarding: Hardware switching is simplified by using VPI/VCI It is connection-oriented, QoS can be guaranteed. Provides traffic control measures. Supports multiple types of services, such as real-time service. ATM was considered to be suitable for all conditions. Some people even raised the idea to create a pure ATM network where the desktop terminal can be reached from the core device. However, this idea was proved to be incorrect. The reasons are as follows: 1. It is too complicated to create the pure ATM network, and the price of the application is too high. 2. The R&D of services fell behind the development of the network. 3. Although ATM switch has been widely used as the core node in the network, the "ATM cell to desktop" service has developed slowly. Background of MPLS Technology Virtual circuit connection (VCC) Virtual path connection (VPC) VC Switching VC Switching VC Switching

7. Background of MPLS Technology • Due to the above mentioned problems, it was inevitable to combine the IP technology and ATM technology to get further development. • MPLS technology was just created by combining the advantages of the switching technology at the core of the network and the advantages of the IP routing technology at the edge of the network. MPLS is created to satisfy the need.

8. Brief Introduction to MPLS Technology Full name: Multi-Protocol Label Switching Functions: MPLS combines high-speed switching of IP and ATM technology. It realizes fast forwarding of IP packets via label switching. Features: Multi-protocol: MPLS technology can support all network layer protocols (e.g. IPV6 and IPX) and link layer protocols (e.g. ATM, FR and PPP etc.) Label Switching: MPLS technology attaches label with fixed length to the packet and replace IP forwarding process with this label.

9. Advantages of MPLS Technology • MPLS provides connection-oriented services for IP network • MPLS provide high quality Internet service • MPLS supports high-bandwidth and high rate IP forwarding • MPLS guarantees QoS and security while providing IP services • MPLS has traffic engineering capability • MPLS supports VPN function

10. Introduction to MPLS Solution MPLS is the short form of Multi Protocol Label Switching. MPLS is a kind of technology between layer 2 and layer 3, i.e. it is a 2.5 layer technology It is a standard routing and switching solution that combines the label switching technology and the layer-3 routing technology. Multi-Protocol means the technology can work together with multiple network protocols. Layer-3 routing is implemented at the edge of MPLS network, and layer-2 switching is implemented inside the MPLS network.

11. Key Terms of MPLS Label The label is an integer identifier with fixed-length. It usually is encapsulated between the layer-2 encapsulated header and layer-3 packet of the data link layer. The label is mapped to the FEC via the binding process. FEC Forwarding Equivalence Class is a group of packets that are processed in equivalence manner during the forwarding process. FEC can be identified via the address, tunnel, COS etc. Usually assigned with the same labels on one device. LSP Label Switching Path: A FEC data stream is attached with specific labels on different nodes and packets are forwarded according to these labels. The path via which data stream flows is LSP. LSR Label Switching Router is the core router of the MPLS network. It is responsible for creating the LSP and initiating the change of next hop. It provides the functions of label switching and label distributing. LER Label Switching Edge Router is mainly responsible for FEC dividing, traffic engineering, LSP initiating, IP packet forwarding, Diff-Serv etc. At the edge of the MPLS network, LER divides the traffics into different FECs and request for labels for the FECs. It provides functions of traffic classification, label mapping and removal of label. LDP Label Distribution Protocol is implemented within the MPLS domain to assign the labels for the devices.

12. Ingress LSP Egress LSRy LDP Protocol LDP LSRx LSRz LERf LERe LERd LERa LERb LERc LDP MPLS Domain MPLS Network Model

13. Operational Principles of MPLS Label Label Label Traditional IP forwarding Traditional IP forwarding MPLS domain Label forwarding The traditional IP forwarding is adopted for the switching out of the MPLS domain. The switching in the MPLS domain is conducted according to the labels, and there is no need to search for the IP.

14. Operational Principles of MPLS The label distribution protocols (e.g. LDP, RSVP etc.) are implemented to assign the corresponding labels for the devices within the MPLS domain. The propagation process of IP packets through the MPLS domain is as follows: 1. The LER at the entrance receives the packets and assigns the corresponding labels for the packets. 2. The backbone LSR receives the labeled packet, searches the label forwarding table and uses a new outgoing label to replace the label in the incoming packet. 3. The LER at the exit receives the labeled packet, deletes the label and performs the traditional layer-3 search for the IP packet.

15. MPLS Label is a 20-bit integer between 0 and 1048575. It is used to identify the FEC. The label is encapsulated between the layer-2 header and the layer-3 data of the packet, and it only has specific meaning for the local domain. MPLS Label

16. Label Stack The label stack consists of two or more MPLS labels. Theoretically, there is no limitation on the label nesting, and it can support all kinds of services. Network layer header closely follows the label whose bottom of stack is set to 1 bit; Packet forwarding is based on the top label of stack. When receiving one packet, LSR will check top label to decide the next hop.

17. Implementation of MPLS In order to implement the MPLS label switching, we should create the label switching path (LSP) in advance. Actually, the LSP is created by assigning the labels for the nodes on the path. The next section describes how to create the LSP.

18. Creation of LSP Method for Creating LSP The following are the 3 methods commonly used to create the LSP (to assign the labels): Data stream driver: the creation of LSP is triggered by the received packets. Topology driver: the creation of LSP is triggered by the topology information (the routing information). Application driver: the creation of LSP is triggered by the application (e.g. QoS). Compared with the data stream driver and the application driver, the topology driver has the following advantages in the label value: The label value setting and label distribution are applied to the control information, it will not cause huge network overhead. The label value setting and label distribution are implemented before the arrival of the data, it will not cause delay. So the topology driver method is usually used for assigning the labels.

19. Creation Process of LSP There are three steps for establishing LSP in MPLS network: After booting up via network, routing protocols (BGP, OSPF and IS-IS etc.) are enabled on nodes to set up routing tables. Under the control of LDP, LIB is set up on each node according to routing table. Map and link incoming labels and outgoing labels on ingress LSR, medium LSR and egress LSR to form one LSR.

20. 47.1 1 3 2 1 3 RC 2 RB 1 47.2 3 47.3 2 RA Step 1: Creation of Routing Table • Routing table is set up on each router under the function of dynamic routing protocol.

21. Mapping: 50 Mapping: 40 Step 2: Creation of LIB 1 47.1 3 2 RC 3 1 RB 1 2 47.3 3 47.2 2 RA

22. IP 47.1.1.1 IP 47.1.1.1 Step 3: Creation of LSP RB 1 47.1 3 3 RC 2 1 1 2 47.3 3 47.2 2 RA

23. RC RB RA Label Stack PHP Mechanism The last hop is assigned with a special label 3.

24. Label Distribution Protocol (LDP) LDP Overview: The Label Distribution Protocol is a protocol for creating the label dynamically. It is based on the UDP (User Datagram Protocol）/TCP（Transmission Control Protocol). The protocol information is transmitted in the hop-by-hop manner according to the routing table. LDP can inform the label switching routers of the Forward Error Correction (FEC) and the mapping relationship of the labels, and at last the label switching path is generated. Through LDP, the FEC is associated with the LSP, and the network prefix traffics are mapped to the LSP. According to the regulation of RFC3036, LDP includes the mechanism of neighbor discovering, label requesting, label deleting, label mapping and error correction.

25. Label Distribution Protocol (LDP) Operational Principle of LDP LSR sets up and maintains the LIB according to the binding information between the label and the FEC. The two LSRs that exchange the FEC/label binding by using the LDP are called as the "LDP Peer". LDP is used by the LSR to bind the FEC and the labels and to inform the neighbor LSR of this binding. In this way, the LSRs will get the consensus about the binding relationships of the received labels.

26. Label Distribution Protocol (LDP) According to the time sequence, the LDP processing includes the following 4 stages: Discovery stage: Discover LDP peers automatically by sending Hello messages to neighboring LSRs periodically; Session Establishment and Maintenance stage: Implement TCP connection and session between LSRs, and conduct initialization (negotiation of parameters); LSP Establishment and Maintenance state: Assign label for FEC to be transmitted between LSRs and establish LSP; Cancel of Session: When session hold-time is out, terminate the session.

27. R1 R2 Neighbor discovery: It is implemented by sending Hello messages mutually (UDP/prot:646/IP:224.0.0.2). Establish TCP connection: The end with larger address initiates connection actively.(TCP/port:646) M Session initiation: The Master router sends initialization message which carries negotiation parameter. M The Slave router checks if the parameter can be accepted. If it can, slave router will send initialization message and carry negotiation parameter. M The Master router checks if the parameter can be accepted. If yes, the Master router will send keepalive message. M After receiving Keepalive message mutually, session is established. M Once received any error message during this period, session will be closed and TCP connection will be disconnected. Establishment and Maintenance of LDP Session

28. Contents • Evolution History of PTN Technology • MPLS technology • MPLS-TP Technology

29. Brief Introduction to MPLS-TP Technology MPLS-TP： MPLS Transport Profile is a kind of connection-oriented packet transport technology. It extends downwards from core network. In April 2008, IETF and ITU-T set up the associated work group named JWT to develop the MPLS-TP technology. IETF was responsible to develop the T-MPLS/MPLS-TP standards and ITU-T was responsible to raise the requirements for transmission. MPLS-TP is based on MPLS technology. Its relevant standards provide complete carrier-level scheme for deploying packet switching transport network. MPLS-TP is based on the IP core network. It simplifies the MPLS/PW technology, removes unnecessary IP functions, and meets the requirement of packet transport. To maintain integrity of point-to-point OAM, MPLS-TP adopts the concept of layered network, OAM and linear protection. It is independent from the client signals and the management network signals. It meets the requirements of the transmission network. MPLS-TP takes full advantage of technical advantages of connection-oriented MPLS technology in QoS, bandwidth sharing, differentiated service and other aspects. Based on the hierarchical network structure of the IP transmission network, MPLS-TP provides the following functions: Support multiple services Connection oriented Robust scalability Carrier-level QoS, bandwidth multiplexing Efficient bandwidth management and traffic engineering Strong OAM and network management functions Provides 50ms protection switchover and recovery Supports dynamic control plane Low CAPEX+OPEX

30. Evolution of MPLS-TP Technology MPLS-TP is a subset of MPLS. Routing and signaling protocol last hop pop-up Label merging Layer-3 functions TDM kernel GFP encapsulation Virtual cascade LCAS Frame structure Label switching Differentiated QoS End-to-end OAM Carrier-level protection switchover Clock synchronization IP/MPLS MPLS-TP SDH • Simplified and discarded parts: • Simplifies the complicated protocol family of MPLS • Simplifies the control plane • Doesn't support PHP • Simplifies the data forwarding plane • Doesn't support label merging • The following aspects of MPLS are adopted: • Frame structure • Label switching principle • Label switching path • Differentiating service (Diff-Serv) • Label space and identifier assignment • TTL processing • New contents added: • End-to-end OAM, which is similar to SDH • Enhanced protection switchover, which is similar to SDH • Linear subnet protection and ring-network protection, supports the APS protocol, and adopts the concept of layered network • Clock synchronization • Increases the depth of the label stack • Supports bi-directional LSP MPLS-TP = MPLS/IP + OAM + Protection

31. MPLS-TP Functions V.S. MPLS Functions

32. Management Plane Management Plane Control Plane Control Plane Data Plane Data Plane Physical Layer Physical Layer Structure of MPLS-TP System • MPLS-TP system contains 3 planes: the data plane, the management plane and the control plane.

33. Structure of MPLS-TP System Data Plane The data plane is responsible to transmit the information from one point to another point in uni-directional or bi-directional manner. It can detect the connection status and provide the result to the control plane, and it also undertakes the transmission of control information and network management information. The major function of the MPLS-TP data plane is to put the different service signals into the MPLS-TP channel and implement the packets forwarding according to the MPLS-TP labels. The main method is to conduct indirect mapping via the MPLS-TP channel or encapsulate the signals into the MPLS-TP transmission tunnels and transmit the data in the packet network. Meanwhile, the data plane also conducts the OAM and protection operations. Management Plane The management plane is responsible to manage the data plane, the control plane and the whole system. It also coordinates the operations between these planes. The functions of the management plane include: performance management, fault management, configuration management, charging management and security management. The MPLS-TP management plane provides the functions of end-to-end in-domain or inter-domain fault management, configuration management, performance management, user management and security management. Control Plane The control plane is made up of a group of control modules that provide specific routing and signaling functions, and it is supported by a signaling network. The information about the interaction and communication between the control plane modules can be acquired via the interfaces. The control plane is mainly responsible to create, release and remove the connection. And it manages the monitoring and maintenance entities.

34. Relationship between Client Layer and Server Layer in MPLS-TP Network The following figure shows the relationship between the client layer and the server layer in MPLS-TP network: • In MPLS-TP network, client signals and control network are completely independent. • Client signals carried by MPLS-TP can be IP/MPLS or Ethernet. That is MPLS-TP exists as the server-layer network of Eth/IPMPLS.

35. MPLS-TP Interfaces The MPLS-TP network provides the User Network Interface (UNI) and Network Node Interface (NNI). NNI can work as both the intra-domain interface of a single management domain and the inter-domain interface between management domains. The interfaces between CE and PE are UNI interfaces, where UNI interface of CE device can be expressed as UNI-C and UNI interface of PE device can be expressed as UNI-N. The internal interfaces between PE and PE are NNI interfaces. The definitions of MPLS-TP network interfaces are shown in the following figure:

36. Hierarchical Structure of MPLS-TP Network MPLS-TP network has the following layers from top to bottom: MPLS-TP channel (TMC) layer, MPLS-TP path (TMP) layer , MPLS-TP section (TMS) layer and transmission medium layer.

37. Hierarchical Structure of MPLS-TP Network TMC layer: It provides point-to-point transport network services, that is to provide client with point-to-point signal transmission. TMC functions as PW layer of PWE3 (or virtual circuit layer). TMP layer: It shows characteristics of point-to-point logical connection. It provides a transport network tunnel and encapsulates one or more client services into one larger tunnel so that transport network can implement more economic and valid transmission, switching, OAM, protection and restoration. TMP functions as tunnel layer in MPLS. TMS layer: It is optional. It indicates virtual connection between neighboring modes. It guarantees the integrity of transmitted data between two nodes at TMP layer, such as SDH, Optical Transport Hierarchy (OTH), Ethernet or wavelength channel. Transmission medium layer: It refers to the transmission media that supports the TMS layer, such as optic fiber, radio and so on. PE P PE Section E1 PWE3 ATM PWE3 Tunnel Ethernet PWE3

38. MPLS-TP Transmission Principle The connection-oriented feature of the MPLS-TP network is realized by the PW technology. By using the PW, the service provider can transmit the traditional circuit-based network services, as well as the new services, in the converged network based on packet technology. Transmission process: The customer edge device CE1 is connected to the provider edge device PE1. PE1 encapsulates the original service data and transmits the packet via the Peat the receiving end, PE2 implements frame verification and sequence re-arrangement and restores the received packets to original service data, and then send the data to the customer edge device CE2.The process is shown in the following figure:

39. MPLS-TP can be considered as a tunnel technology based on MPLS labels. It uses a group of MPLS labels to identify the end-to-end forwarding path (LSP). MPLS-TP has two layers. The inner layer is the PW layer that identifies the type of the service; The outer layer is the tunnel layer that identifies the forwarding path of the service. The tunnel is the end-to-end label forwarding tunnel based on MPLS-TP. The local packets are encapsulated as PW PDU via the PW and transmitted via the tunnel. The service data are encapsulated/de-capsulated by the PE device, then they are restored to the local format and sent to the destination CE. Emulation service, such as TDM, ATM Emulation service, such as TDM, ATM Emulation service Payload encapsulation PW Payload encapsulation PW multiplexing PSN tunnel PSN Physical layer PW multiplexing PSN tunnel PSN Physical layer PSN tunnel Packet Forwarding Technology

40. B-DA: the 6-byte destination MAC of the Ethernet encapsulation （Tunnel_L determines the forwarding path, B-DA is the MAC address of the next hop node) B-SA: the 6-byte source MAC of the Ethernet encapsulation 0x8100: the 2-byte identifier of the Ethernet data frame B-VID: the 2-byte outer VLAN tag 0x8847: the 2-byte "Pw Over MPLS-TP" identifier Tunnel_L: the 4-byte tunnel Label (TMP) PW_L: the 4-byte pseudo wire label (TMC) CW: the 4-byte PW control words Customer Frame: the user data, payload (the user VLAN may be included) The user packet should be added with 26 bytes in total. Customer Frame CW PW_L tunnel_L 0x8847 B-VID 0x8100 B-SA B-DA Packet Encapsulation

41. Pnode P node PE node CE node PE node CE node Packet Forwarding Process Customer Frame Customer Frame Customer Frame Customer Frame Customer Frame Customer Frame Peel off label Add label Reserve PW label Reserve PW label Reserve PW label CW CW CW CW PW_L PW_L PW_L PW_L tunnel_L tunnel_L tunnel_L tunnel_L 0x8847 0x8847 0x8847 0x8847 Switching tunnel label Switching tunnel label Switching tunnel label B-VID B-VID B-VID B-VID 0x8100 0x8100 0x8100 0x8100 Update next hop MAC Update next hop MAC Update next hop MAC B-SA B-SA B-SA B-SA B-DA B-DA B-DA B-DA

42. 45 8847 MAC1 MAC2 Sample of Packet Forwarding Process MAC 1 MAC 2 MAC 3 MAC 4 1 1 3 2 4 1 1 3 4 2

43. 43 8847 MAC2 MAC3 Sample of Packet Forwarding Process MAC 1 MAC 2 MAC 3 MAC 4 1 1 3 2 4 1 1 3 2 4

44. 43 43 8847 8847 MAC2 MAC2 MAC3 MAC3 Sample of Packet Forwarding Process MAC 1 MAC 2 MAC 3 MAC 4 1 1 3 2 4 1 1 3 1 3 2 4 2 4

45. 43 43 8847 8847 MAC3 MAC2 MAC4 MAC3 Sample of Packet Forwarding Process MAC 1 MAC 2 MAC 3 MAC 4 1 1 3 2 4 1 1 3 3 2 4 2 4