Analysis of Software Streaming Data

In this paper, we analyze the software streaming data for stable service and efficient resource allocation of the server. We introduce the software streaming service and the message flow of the service. Based on the message flow, we firstly divide the software stream into the initial phase and the on-demand phase. And we further divide the initial phase into the downloading period and the processing period. Finally, we classify software into the downloading based software and the processing based software according to the analysis of the proportion of each period. And we also define two external environmental factors (current available network bandwidth between server and client and the processing power of the client PC) which could affect overall performance of the software streaming service. We demonstrate the external environment that affects the ALT and SRU. So we validate it by several experiments. The result shows that the streaming server consumes the most resource when the rich client uses the downloading based software in the high speed network, and there is an inverse proportion between the ALT and the SRM. Using all the information (software type, network condition, client processing power), the streaming server can allocate resource to a client more accurately.