A Social Aware Approach for Online Social Networks Data Allocation and Replication

摘要

Data in Cloud Computing should be stored and replicated across different geo-located data centers for high data availability. Facebook, Twitter and many other social media are examples of booming Cloud-based web applications; known as Online Social Networks (OSN). These OSN services started in the first decade of the twentieth century and with an exponential increase of the amount of shared data and number of users. Data storage and replication for this massive amount of data is a hot research topic especially for social networks. In social networks, the user writes her/his data while most of the reads is done by her/his active friends; not by himself as usually happens in other web applications. Each OSN has several datacenters geo-distributed across the globe to provide the social media service for the worldwide users. In this paper, we propose a novel data storage and replica algorithms across a group of cloud data-centers based on the friendship matrix and their spatial distribution. The extensive simulation studies that have been conducted to compare the proposed algorithm with the famous Cassandra rack unaware algorithm, which is currently used in Facebook and Twitter, report the outstanding performance of the proposed algorithm in terms of decreasing the average user's access delay while preserving an acceptable load balance.