Self-Similar Traffic Checker

The workload of a system is rarely uniform. Your particular system may appear to have its own signature fluctuations. However it has been demonstrated that most workloads, however diverse they may seem, fall into a few basic patterns (statistical probability distributions) as a law of nature. For example, in traditional client server systems the request arrivals follow a Poisson distribution. This pattern is extremely common and applies very accurately to several systems such as telephone calls at a call center or customers at a Pizza Hut. With such distributions the mean value is sufficient to characterize the workload completely.

Turning now to Internet facing systems arrival pattern appears to be chaotic and difficult to model. It has been determined from empirical and theoretical research that the incoming traffic will follow a self-similar pattern. The self-similar or fractal nature means that the pattern repeats itself over different dimensions of time. For the purpose of modeling the traffic, the degree of randomness or burstiness of the traffic is considered. For more details refer to Methodology Pack WWM851.

Self-similar traffic can be checked using the Variance Time Plot (VTP). As we aggregate self-similar traffic at coarser and coarser time scales, the variance decreases (decays). VTP analysis explores and quantifies this decaying variance. For more details refer to Methodology Pack WWM851.

Related Tool: Self-Similar Traffic Generator

Instructions to use the tool: In the textbox provided below, please enter the time series data and separate them with a space or a comma. For example each value could be the number of hits per second taken for a finite period of time. The minimum number of data points required is 16.