There are a lot of products and services available in today's marketplace that make the claim of managing end-user performance both on an enterprise network and to Internet web sites. Many such products simply take particular metrics and report them, leaving the engineers and technicians the job of uncovering the location, cause, and potential remedies for service degradation. Services, like those of Keynote Systems and Exodus Communications, which measure user response times from different geographic locales, are useful when comparing one company's performance to another but they still lack the element of diagnosis.
This paper will deal with the aspects of end-user performance that are important to the technicians that must manage the networks as well as the engineers who are responsible for designing and improving the network infrastructure. Although certain products will be mentioned by name, the intent is to highlight functionality rather than serve as a product endorsement. Care should be taken by the reader to understand that, while this writer has certain preferences and prejudices regarding industry products and services, there are a lot of offerings available to meet the needs defined here.
.
The first aspect of this paper will deal with the proper performance testing and benchmarking of network infrastructure hardware devices. Industry standard practices and guidelines are the focus for the discussion, with enough information given to stimulate ideas for testing in a production environment. With this as a foundation, the next logical step is to consider performance testing from the end-user perspective. This latter discipline will offer a suggested methodology for isolating service degradation to or from the network, which is of the utmost importance and the major point of this paper. The basic argument could be stated simply as, "there is little practical use in measuring end-user performance over the network when the measurements fail to offer a solution.
