Introduction
Since data traffic started dominating telecom networks there was a need for transport data networks as opposed to transport TDM networks. Traditional transport technologies like SONET, SDH and OTN are good for voice traffic. There was a need for new technology that will continue the useful features of transport technologies and packet switched in nature.
MPLS-TP (Multi Protocol Label Switching – Transport Profile) is an effort by IEEE & ITU-T to have a technology that meets all the requirements of transport networks but is packet switched in nature.
Connection Oriented Transport
Major motivation for developing MPLS-TP was the need for the circuits in Packet Transport Networks. Usually packet transport switches each packet individually. In connection oriented transport a connection is first setup between the end points and the traffic follows that path through the network. This makes the Packet Transport Network similar to the TDM networks and streamlined management and migration of the transport network.
The concept of Label Switched Paths or LSPs from MPLS technology is already tested and successful in the inter-networking world. It made sense to adapt it for use in Packet Transport Networks. Yet there was a need to simplify the working of MPLS to make it more suitable for use in the Packet Transport World. For that, some features removed from the traditional MPLS, since it felt that these were not needed in Transport World and would only the network.
The features from MPLS that are not supported by MPLS-TP are:
The features from MPLS that are not supported by MPLS-TP are:
- MPLS Control Plane
MPLS-TP does not need LDP or any other control plane protocol to set up the circuits. Instead a user provisioned model followed. The user can provision a circuit from a centralized Network Management System (NMS) in a way similar to TDM networks.
- Penultimate Hop Popping (PHP)
PHP is use by MPLS Edge Routers to reduce the load of two label lookups. But this causes problems with QoS and disabled in MPLS-TP.
- LSP Merge
Merging two LSPs (going to the same destination) reduces the number of labels used in the network. But it makes it impossible to differentiate between traffic common from two different sources before the merging happened. To simplify things in transport networks, LSP merge was also disabled.
- Equal Cost Multi Path
In traditional IP/MPLS networks different packets between a source-destination pair can take different.
The introduction of circuits is the main reasons to adding a lot of functionality to transport networks.
OAM
Operations, Administration and Maintenance have always been important for transport networks. And that is one of the mail reasons for the success of SONET/SDH. Recent developments have brought this functionality to Packet Transport as well. Standards like IEEE 802.1ag, ITU-T Y.1731, Bidirectional Forwarding Detection (BFD), LSP-Ping, and LSP Traceroute are enabling OAM features on Packet Transport.
Alarms can rise on certain events/failures. Loopbacks can used to localize faults and end-to-end connection parameters like throughput, packet loss, latency and latency variation (jitter). This can measure and continually reported. The OAM can apply across multiple layers across the network. For eg, an End-to-End connection can monitor and also for a particular network operator or a particular sub-net, paths. This is especially true when multiple equal cost paths exist. But this is in conflict with the concept of a circuit where all the traffic should follow the same path. Hence ECMP disabled.
Linear Protection
Traditional SDH has offered end-to-end Path protection and that too within 50ms. The same is required from Packet Transport. Connection oriented Packet Transport concept with OAM on those circuits makes it possible to detect the faults and switch the traffic from working to protecting path fast. These protect paths would be pre-provisioned and kept ready, just like in SDH. This functionality has been standardized by ITU-TG.8031 standard.
It is possible to define the protection as revertive or non-revertive and option for the user to give forced switch or manual switch command. It is also possible to provide Linear Protection in a stitched pseudowire.
Quality of Service
In a packet switched network since all the packets share the same infrastructure, Quality of Service and prioritization are important. A high priority or delay critical packet should not be held up in a queue by low priority or non-premium traffic. Hence all packet switched networks need QoS mechanisms.
MPLS & MPLS-TP label has a 3 bit Traffic Class (formerly called EXP bits). These can support 8 classes of service per LSP. At the LERs the IP or TCP header fields can use to mark the TC bits or it can be statically provisioned by the service provider. With QoS, both a flat QoS and a hierarchical QoS model can be supported as shown below. Hierarchical QoS can ensure fairness of service between different services, between different customers and across different tunnels.
Flat QoS. All the traffic will be prioritized based Flat QoS on PCP/IP DSCP field
Hierarchical QoS:
3 Levels of QoS
- First level scheduler ensures that each CoS within a service gets the promised bandwidth.
- Second level scheduler ensures that service (pseudo-wire) gets the required priority within a tunnel.
- Third level scheduler ensures that each tunnel get the requisite priority within the egress port.
Scalability
MPLS-TP follows a label similar to MPLS. The label has 20 bits for label value which allows for 1 million unique labels. Also, label stacking allows multiple levels of hierarchy to be created. There’s no limit on the number of levels that can be stacked. This makes MPLS-TP, just like MPLS highly scalable. One limitation of MPLS was that the entire database of network topology had to be present on each router. Since each router had limited memory and processing power, they could handle only a particular network size. In MPLS-TP by moving all this intelligence to a central NMS, so high-capacity multi-processor servers can be deployed to do this processing, but the individual nodes become simpler and cheaper.
Summary
MPLS Transport Profile (MPLS-TP) is a profile of MPLS. It will be designed for use as a network layer technology in transport network.
This MPLS profile is simplified for transport network with some of the MPLS functions turned off (e.g. PHP, LSP merging...) and with enhancement based on transport network requirements (e.g. In-band Carrier Grade OAM).
This MPLS profile is simplified for transport network with some of the MPLS functions turned off (e.g. PHP, LSP merging...) and with enhancement based on transport network requirements (e.g. In-band Carrier Grade OAM).
- It is a connection-oriented packet-switched application.
- Its design and standardization effort for MPLS-TP is being progressed in a cooperation between ITU-T and IETF.
- The required protocol extensions to MPLS being designed by the IETF based on requirements provided by service providers.
- Solution is based on existing LSP and Pseudowire constructs.
- MPLS-TP tunnels provide the transport network service layer over which IP and MPLS traffic pass through.
- MPLS-TP tunnels help the transition from SONET/SDH and TDM technologies to packet switching to support services with high bandwidth utilization and lower cost.
- Transport networks are connection-oriented, statically provisioned, and have long-lived connections.
- Transport networks usually avoid control protocols that change identifiers (like labels).
- MPLS-TP tunnels provide this functionality through statically provisioned bidirectional label switched paths (LSPs)
This is one of the ideal technologies to build packet transport networks of the future.








