Please note that this package measures the net performance from the viewpoint of an end user. This implies that, using only these data, on beforehand no conclusions can be drawn about the performance of one particular network in the chain of providers between sites participating in the measurements. The results we see are a cumulative effect of all intermediate networks. However, there is a possibility to obtain detailed information about a network, but this would imply installation of the test scripts at strategic sites of the network provider, such that a particular route is isolated.
From our application oriented viewpoint we are mainly interested in the end-to-end results of the provided network services and their inter-connections. One may interpret this as not very fundamental, and therefore have the opinion that the monitor programs of the network providers give sufficient information itself. Although networks are well capable to monitor the performance of their own connections, we believe that they are less focussed to the overall performance to the route as a whole, and their possibility to do so is also rather limited. We think that this is the interest point from a tools as ours: to match the performance figures of the networks with end-user measurements. Because our tool requires some machine power and net capacity we agree that limiting the monitoring to a fixed time interval is preferable above a continuous monitor.
Combined for all n sites, the total throughput measurement time Tall becomes:
For the typical situation of six involved sites and an individual throughput duration of 10 seconds this implies per site a total throughput measurement time of 50 seconds. When the measurements are also typically repeated each hour the relative duration of the tests is 1.4 %.
The implications for the network load depends from its topology. Assume the topology of the WAN between the six sites A, ..., F shown in figure 1 below. Assume also that the traffic is routed as follows:
80 s 80 s A ---------- B ---------- C | | | | | | 53 s | | 73 s | 53 s | | | | | | F ---------- E ---------- D 80 s 80 s
This technical document describes the calculation of the load numbers in figure 1.