Cuckoo: towards decentralized, socio-aware online microblogging services and data measurements.
Tianyin Xu,Yang Chen,Jin Zhao,Xiaoming Fu
DOI: https://doi.org/10.1145/1834616.1834622
2010-01-01
Abstract:ABSTRACTOnline microblogging services, as exemplified by Twitter [9] and Yammer [12], have become immensely popular during the latest three years. Twitter, the most successful microblogging service, has attracted more than 41.7 million users as of July 2009 [25] and is still growing fast. However, current microblogging systems severely suffer from performance bottlenecks and central points of failure due to their centralized architecture. Thus, centralized microblogging systems may threaten the scalability, reliability, as well as availability of the offered services, not to mention the extremely high operational and maintenance cost. However, it is not trivial to decentralize microblogging services in a peer-to-peer fashion. The challenges first derive from the heterogeneity of the inherent online social network (OSN) features. The non-reciprocation feature of microblogging services also increases the heterogeneity. Moreover, different from traditional approaches used in centralized server-based systems, an efficient, robust and scalable approach for data collection and dissemination in such distributed heterogeneous environments is desirable. In this paper, we present a decentralized, socio-aware microblogging system named Cuckoo. The design takes advantages of the inherent social relationships while leverages P2P techniques towards scalable, reliable microblogging services. Besides, Cuckoo provides a flexible interface for data collection while circumventing unnecessary traffic on the server. We discuss the benefits that our system may bring for both service providers and end users. We also discuss the technical aspects to be considered and report our work in progress.