Optimized Delivery Mechanisms for QoS of Traps

With the growing reliance on telecommunication, it is of paramount importance that the telecommunication network operators maintain high availability and low mean time to repair (MTTR). To achieve this objective at a low operational cost, central network management becomes mandatory. The elements of the network use traps to convey any problem to the central management station. Often there is a possibility that a failing network element (NE) floods a management station with redundant traps and congests all available network bandwidth. When there is a flood of traps, the manager gets overloaded and critical traps might get lost or delayed. The analysis of trap data from multiple corDECT telecom networks confirm that a major portion of traps during a flood is contributed by a few repetitive events from a small number of NEs. Operator should be presented only with the actual faults without repetition. Methods are needed to ensure reliable and timely delivery of traps in this distributed network.

We propose a correlation engine that correlates and removes the repetitive and cleared traps that are received at the manager so as to present only the actual faults to the operator. We propose a technique by which the manager can control the NEs that are flooding it with traps. Apart from this, to avoid forwarding of short lived and redundant traps we propose distributed correlation and compression techniques at the NEs. These techniques put together helps reduce the total number of traps delivered and give control to the management station to maintain QoS for trap delivery. This also ensures that the management station is not overloaded.

The proposed rule-based correlation is implemented and is being used in all corDECT deployments. It is seen that the correlation engine is very effective in reducing the number of traps presented to the operator. The other proposed techniques are implemented in a corDECT network both at the NMS manager and device side. Using experiments in a laboratory setup various parameters for the scheduling algorithm are tuned to obtain optimal results. The field trace driven experiments show that there is considerable gain, in terms of control at the manager, number of events reported to manager and timely trap delivery, even under heavy flow of traps.