A case study on oscillating behavior of end-to-end network latency

Understanding network latency is important for providing consistent and acceptable levels of servicesin network-based applications. Due to the difficulty of estimating applications' network demands and the difficulty of predicting network load, however, the management of network resources has often been ignored in network-based systems. This paper presents network traffic oscillating behavior that has been observed in real operational networks. The basic idea on the study is that a variation of network latency is strongly correlated with the past history of the latency. Four typical network traffic status are defined based on the stability and the burstiness of latencies. Observations of network latencyare an open research area across multi-time scale levels and the proposed network status would be helpful to simplify issues in the area.