Distributed B-Tree with Weak Consistency [chapter]

Gregor V. Bochmann, Shah Asaduzzaman
2013 Lecture Notes in Computer Science  
B-tree is a widely used data-structure indexing data for efficient retrieval. Highly parallel operations are desired by modern-day cloud computing platforms on high-volume and highly dynamic sets of data. This motivates decentralized indexing structures for data-organization that avoid any performance bottleneck. In a decentralized B-tree, parts of the structure are distributed among different processors and some parts are replicated. Maintaining strong consistency among the replicated states
more » ... distributed data-structures has been found problematic in large-scale cloud platforms. We show in this paper that data retrieval can be performed correctly with much weaker consistency criteria among the replicated parts of the distributed B-tree. To accommodate the dynamic changes concerning the data insertion/deletion and the distribution of retrieval load, the state of the B-tree is updated by splitting and merging tree-nodes. We provide update algorithms that work under the weak consistency criteria and maintain them. The update operations are initiated independently by individual processors and modify the states of only a few processors in their immediate neighbourhood.
doi:10.1007/978-3-642-40148-0_12 fatcat:l5cst2i5uzhvjou4bexgtjnagq