The problem with analysis of the performance statistics of computers in a network is the fact that each statistic is usually a numerical value that varies with time, and there are quite often a significant number of these statistics. Another problem with analysis and interpretation of these statistics is the relationship between these statistics. In many situations there is a causal relationship between the values the various statistics take, the relationship being very difficult to discover.
The key to interpretation of data, according to [Robe91], is the selection of an appropriate representation scheme. Given the time-variant nature of the data involved here animation and other graphical schemes represent a good choice. It is this approach which NetSpy takes in presenting the statistics to the user.
Visualization of the statistics is performed using 3D graphics and a user, typically a network administrator, has the freedom to select which statistics to monitor, and which of several visual controls to use.
Existing systems, such as SeeNet ([Beck95]) and Hy+ ([Cons93]), make extensive use of 2D graphical displays to visualize information about the performance of a network. In most cases the performance of the individual nodes is not considered outside the context of network traffic. While these tools are very strong in the area of network analysis, they do have this shortcoming.
The network tools alluded to above are quite mature and present their information to the users in ways which are easy them to understand. A limitation of these tools however is that they only display information about the network itself, not allowing for visualization of statistics about the nodes outside of the context of network traffic. This is a significant shortcoming given that the use of the computers for other services is on the increase. For example, a single computer could be used as the host for large database system, which is accessed by other computers on the network. A bottleneck in throughput of the database could be due to congestion on the network, or due to excessive loading on the computer hosting the database. The exact cause could be quite difficult to locate if it is not possible to monitor aspects of that computer other than just network traffic.
It is this particular shortcoming that NetSpy tries to address. It presents a toolkit which an administrator can use to create visualizations of any statistic that can be collected by the supported collection services. It is this flexibility of what is collected and visualized that is the strength of the NetSpy system.