MPLS Traffic Engineering (TE)

Traffic engineering refers to the process of selecting LS paths chosen by data traffic in order to balance the load on various links, routers, and switches in the network. This is most important in networks where multiple parallel or alternate paths are available. The goal of Traffic Engineering is to facilitate efficient and reliable IP network operations while simultaneously optimizing resource utilization and network performance. Prior to MPLS TE, this technique is possible with either IP or ATM depending on the protocol used between a pair of edge routers in a network.

MPLS Benefits in Traffic Engineering

Traffic Engineering in MPLS involves the technique of directing traffic that flows within a network. Several routing procedures implement packet forwarding for a secure transmission. The following advantages enhance traffic engineering:

• Minimize network congestion: An MPLS network can implement TE to curtail network blockage and boost up the performance. All routing techniques in use are modified to map packet data to network resources. Such a mapping process can handle bottlenecks of packet overcrowding with suppression of latency, jitter, and loss factors. MPLS TE allows exploitation of bandwidth in use rather than allocating new bandwidth to operate traffic engineering. The tunnels direct the traffic from congested path to under-utilized path available to alleviate traffic congestion.
• MPLS Fast Reroute for link/node failure: MPLS Fast Reroute functionality handles link or node failures by directing encapsulated traffic to a preconfigured secondary path when the primary one fails. This is not possible in case of IP networks, as redirecting mechanism is not applicable here. The highlight is MPLS assures highest reliability and network uptime with appropriate mechanisms to recover from network congestion and other bottlenecks.
• Deployment flexibility: A TE system is efficient even when the MPLS network implementation is under-developed. Any combination of circuits with T1, T3, optical carriers, or Ethernet can be assimilated into a MPLS setup. Offices with multiple branches over the globe make the most of this deployment flexibility with different combination of connections. It is flexible during situations when the overflowing packets from links are transferred to the available links. MPLS tunnels can also implement traffic engineering without LDP.
• Class of Service (CoS): This 3-bit field determines the CoS value, based on which the traffic in its priority queue for is used for transmission. At the ingress edge, the arriving IP packet is marked with CoS value and they are encoded for reference in MPLS header. This provides fast packet transmission between nodes to avoid network congestion. Functions of CoS are Committed Access Rate (CAR), Weighted Random Early Detection (WRED), and Weighted Fair Queuing (WFQ). Each service class implement traffic engineering by classifying traffic based on available bandwidth in links, manage packet overflow in edge routers, drop probability and network traffic control using algorithms (like round-robin).
• Customer traffic identification: MPLS traffic engineering classifies the customer traffic based on the service provider used in MPLS network. This is solely due to CoS feature of the technology that does traffic categorization.

Limitations of Traffic Engineering

While MPLS renders several gains over its implementation, few setbacks exist based on the network in use and available bandwidth. Find below the technical downsides:

• Over-utilization of secondary links: At the instances of link failures, Fast Reroute of MPLS TE uses backup tunnels to reroute the traffic over the secondary link. While this is a configured setup for recovery, the performance is complete only if those tunnels comprise sufficient bandwidth. A satisfactory level of free capacity is necessary for smooth functioning. In spite of this backup method, frequent failures of network nodes will lead to constant traffic congestion on alternative paths reducing its efficiency overall.
• Manual path setup: To implement traffic engineering, paths require manual configuration irrespective of the presence of Internet Protocol for packet routing. Physical path computation is possible by denoting the consecutive hops occurring from source to destination path. However, this manual setting needs professional solution providers to put manual path setup into practice. In addition, if intermediary nodes are not configured manually, then MPLS TE traffic does not gain importance is treated same as of the regular IP or MPLS traffic.
• Protocol dependency for automatic rerouting: If a network uses OSPF routing protocol (Open Shortest Path First), then automatic path calculation and systematic rerouting of IP traffic can be done in MPLS TE paths. This even applies to networks using IS-IS protocol (Intermediate System to Intermediate System). The existence of LDP in MPLS VPN is essential for tunneling process. Added further, the Area Border Routers (ABRs) require manual identification in a TE path and hence repeated optimization is not feasible once specified. Based on the configuration setting in IS-IS protocol, TE tunneling is set static or dynamic.
• Performance variation in MPLS Fast reroute: Quality-of-Service mechanisms maintain bandwidth for the tunnels that operate as a backup. As intermediate nodes in TE path lack manual configuration, traffic using Fast Reroute on the alternative path will stumble over link failures. Moreover, configuring intermediate nodes in MPLS network manually is not achievable as the technology lacks such options.
• Lack of systematic mapping system: The dynamic mapping of IP traffic on MPLS TE paths is not achievable, as the routers that set up path do not recognize the topology of next OSPF locale.

Request for a free white paper on “Implementation of Traffic Engineering over MPLS & Other Networks”