Analyzing the resource requirements of usage statistics gathering on online newspapers

Web analytics is a process through which website usage statistics and user behavior data are gathered. Such an analytics program can be used as a tool to get anonym information about users, for example who they are, from which site are they are coming, how often they are visiting the website, how much time do they spent on a site, etc. In this paper we focus on the performance of large websites from the perspective of web analytics and the required extra storage space. The number of page views can be millions per day on large websites, thus the stored user-data number (the amount of analytics data generated by a user visit event) is millions per day as well. Poor system design, even related to gathering user behavior statistics, can lead to system crash. Increasing the performance of the hardware exponentially is not possible in the practice, therefore only optimal data processing can lead to efficient system operation. In this paper we show how to calculate the number of userdata records generated in the database per day and total from system startup in general. Based on that we propose a model to calculate the cost of article high score list which is usually generated at least daily in such systems. Finally, the affect of different search engines are examined.