SoMed: A Hybrid DHT Framework towards Scalable Decentralized Microblogging Services

Abstract

Microblogging services have experienced unprecedented growth. They serve more as news media outlets than online social networks, which can predict the H1N1 virus emergency, detect earthquakes, and live major events. These kinds of burst events will cause peak access. Servers will suffer from large amount of concurrent access and lead to single point of failures. Furthermore, divergent behaviors of two categories of users, social users and media users, will lead to asymmetric workloads. For some media users, such as Lady Gaga, who accounts for over 20 million followers, will generate a huge amount of traffic and make the centralized mechanism hard to scale. To address these problems together, we present SoMed, a hybrid DHT framework towards scalable, decentralized microblogging services. We also conduct a simulation with real traces of Twitter. The results show significantly bandwidth saving and response time reducing with high reliability and availability.