Traffic engineering

Traffic engineering is actually a series of tools and methods, whether in network equipment and transmission lines normal or failure of cases, it can from the given infrastructure to extract the best services. That is to say it to install resources is optimized. In fact, it is to the network engineering or network planning a supplement and perfect measures. Traffic engineering trying to make actual network turnover in the most optimal way exists in the physical network among. The IP network, traffic engineering problems discussed are: solve due to traffic congestion, avoid the invalid mapping and resources caused some network resources utilization, and other resources transition is using insufficient contradiction, According to the network topology, rapid, accurate, effectively dynamic redistribution, especially in network traffic occurs when a fault lines or equipment must be more so.

Traffic engineering problems to solve

Intelligent connection

In order to realize the optimization of network resources, reduce the burden of management personnel, MPLS node will adopt signaling protocol establish end-to-end label switching path (LSR). Intelligent connection can be decided by the user (explicit routing), also can realize by pathways computing algorithm. The currently used link-state database can be OSPF - TE (OSPF traffic engineering) or IS - IS - TE (IS - IS traffic engineering) library, but computing algorithm IS the same. Here must be explicitly note, no additional restrictions will undoubtedly in access computing algorithm minimum cost path. If no additional constraints, path algorithm can obviously improve the management personnel allocation, but clearly routing workload to improve turnover distribution network does not pose a substantial contribution.

Bandwidth estimate

In order to improve the path calculation algorithm distribution, but also need the effectiveness of portfolio other parameters. Estimate bandwidth is one of them. In fact, every LSP needs a estimated bandwidth, and each of the link will have a link congestion coefficient, and the coefficient is according to the link capacity and have taken up LSP capacity and valuations calculated results. When adding a new path, each of the relative path to have to calculate its path congestion coefficient. The new link will choose to have the lowest link congestion coefficient of composition.

Bandwidth measurement

Although estimates bandwidth to improve the network resources utilization has greatly improved, but it is still an estimate. Obviously, to the actual portfolio, the more accurate estimate of the route choice the availability is higher. Due to the foregoing path calculation algorithm USES is a kind of static way without considering the actual time-varying link utilities, so the likely outcome of a link volume estimated too high or too low. Low flow estimates that could cause actual link volume, causing we wish to avoid overload of congestion and packet loss; While very high flow estimates that may cause actual link caused by insufficient, other parts unnecessary potential network congestion.

So, let estimation function better bandwidth algorithm is adopted to measure bandwidth. Regular measurement link utilization rate, utilization rate fluctuation, buffer occupy and buffer occupancy rate fluctuation will be vital. These new parameters by routing protocol regularly publish or above a certain threshold in triggered when the release. Use this to dynamically calculating polymerization portfolio, the effective bandwidth. This method than the ordinary bandwidth estimate more precise and thus enhance the network resources utilization.

Network elastic

Current Internet service for enterprise is critical to the success of, so the user need different service usability. MPLS control mechanism should be able to offer lines or equipment failure restore function. The traditional approach is to adopt an overtime detection, namely a passive way, the new method should be adopted active way, the fault early detection and prediction. Low-level failure information with routing and signaling layer should have direct connection, so that more early to trigger restore countermeasures.

Connection priority

Make certain connection than other link has higher priority is obviously one universal demand, so in the LSR must have this MPLS ability, including the connection setup priority and releasing attachment priority. It is worth noting in fault condition of priority to shorten processing high priority connection of recovery time is very important. In a circuitous route of place, because the network breakdown of dynamic, best circuitous route often only at the last minute, therefore connected to determine the priority must dynamic configuration.

Network reorganization rules

In modern communications network, network lines or equipment, a failure will affect hundreds, even thousands of LSP. Disordered release, restructuring routing and second signaling will lead to switch control system overload, causing network reorganization time is too long. Therefore, in a MPLS must have a refined reorganization mechanism, then must specify different priority LSP group restructuring strategies. The advantage of this also is able to quickly will reshuffled network resources occupation information spread to MPLS node, ensure related restructuring is still optimize network resource considerations.

Marker stack

MPLS mark stacks ability can be significantly improved network core recovery time. Such as a portfolio in the core network polymerization, by adding another layer tags can reduce the number of core network LSP. Once the network core fault happens, just click on the small number of LSP routed reorganization, simplified the restructuring task.

Recovery recovery/path optimization

When the fault MPLS network, network equipment recovery after they still should be restored to the original optimal allocation of resources condition. In connectionless networks, portfolio will automatically restored to the shortest path surface, but in connection-oriented nets, must adopt a kind of similar mechanisms, restore the recovery/path optimization is a connectionless networks that often USES in method. MPLS network in LSR can regularly check whether exist than existing LSP better path, if there are, the old LSP should switching to new LSP. A typical example is in network fault recovery after, temporary LSP and return to the original LSP, another kind of circumstance to restore to the specified path.

Signaling performance

Although MPLS is a kind of topological driven network, most LSP existence time is long, for signaling of performance requirements are relatively low. But because the MPLS purpose is as a backbone technology, must quickly effectively to fault recovery, so actually, signaling the performance of MPLS traffic engineering level of automation have very important influence.

Network service quality

MPLS traffic engineering need to provide the service quality, specifically in the light of different service, especially ahead of the multimedia services offer optional/controllable network service quality. This requires each MPLS can LSR of selected LSP distribution specific buffer and scheduling priority, and obligate relevant network resources. Obviously, a dynamic more service network service quality optimizing system for MPLS successfully is very important.