User Profile and Workload Analysis for Local Area Networks

Abstract

Performance analysis tools for computer networks need accurate and comprehensive estimates of user workload. An approach is presented that estimates network impact for a wide variety of end user types and applications that are typical on local area networks. Fourteen user types and nine generic application types are defined, and data is collected to determine the average network bandwidth needed to accommodate the output of individual and aggregate user/application combinations. Workload is estimated using a combination of data obtained from live test experiments, and data collected from the literature. Finally, the implementation of this data in a highly interactive network modeling tool (NetMod) is illustrated with screen images generated during tool execution.