As a result of the development of net technologies and high-speed channels of data translation, new net services and systems have emerged; a new orientation in the field of net technologies & service creation has been established, namely, functional programming on the net service level. Functional programming technology is used to create an effective system for the rapid search and extraction of necessary data from vast net stores, to streamline the data and to develop decisions with the help of appropriate net service resources. Based on the above, a massive industry for data extraction has been developed. The success of these technologies in the net service market is derived from the effective organization of such systems. It should be noted that the effectivity of net services (systems for data fusion, support for decision making, etc), and the initial Quality of Service (QoS) depends on the parameters and properties of the net environment. The net environment has an essential influence on request receiving processes, i.e. on incoming traffic & on the serveice of replying files. Net environments define important characteristics of incoming traffic, such as its nature, interval distribution between incoming requests and their intercorrelation. And for replying process important characteristics are net environment influence on time for data translation by channels and channels loading. All of which are the subject of research on the influence of net traffic properties and parameters of net services characteristics, as well as the development of new models and methods for data translation and net channel resource division. This paper presents a decompositional approach to the modeling and analysis of characteristics of Internet service performance, with consideration to access line loading and different methods of their throughput distribution. The approach in based on the functional-time division of request processing in Internet services. As a mathematical model of request transfers on access lines, the combined queuing models Mn|Gn|1|N + MmGm|K|0 are considered, with registration of different restriction options on buffer capacity and on the number of requests in the system, request service time dependence on length, and different strategies for the division of system resources between request streams. In the first subsystem of the model the requests are processed by the Resource Sharing protocol, and in the second subsystem, by a K access line without waiting. The characteristics of Internet service performance with finite & infinite sources of requests, different strategies of resource access line division, and the distributed function of request length and service time, are investigated and defined.