| Abstrakt: |
The growing significance of the usage of different social networking services in a personal and enterprise context provoked the interest of many researchers to experiment with Educational Social Networks. The standard approach to building an Educational Social Network is using a generic open source solution (like Elgg or BuddyPress) and customizing the user interface and the login system to fit the specific needs. Following such a deployment at the University of National and World Economy (which has a large potential user base - around 20,000 students and more than 500 lecturers), experiments have shown that the classic normalized relational database model of Elgg is not well suited for a significant growth in the user base, the connections and the shared content (the social graph). This paper presents a new dual-layer data model, which attempts to provide a manageable vertical and horizontal scalability. The model is evaluated by comparing the latency of the most critical (for performance, memory consumption and educational usage) operations, both as the social graph grows and with the corresponding values in the original solution. The underlying causes of the differences are analyzed along with the accepted trade-offs. [ABSTRACT FROM AUTHOR] |
