Beehive: A Framework for Graph Data Analytics on Cloud Computing Platforms

Beehive is a parallel programming framework designed for cluster-based computing environments in cloud data centers. It is specifically targeted for graph data analysis problems. The Beehive framework provides the abstraction of key-value based global object storage, which is maintained in memory of...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Proceedings of the International Conference on Parallel Processing s. 331 - 338
Hlavní autoři: Tripathi, Anand, Padhye, Vinit, Sunkara, Tara Sasank
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.09.2014
Témata:
Cloud computing
Computational modeling
Data analysis
Data models
Graph algorithms
Instruction sets
Optimistic concurrency control
Parallel processing
Parallel programming
Programming
Servers
Transactional memory
ISSN:0190-3918
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Beehive is a parallel programming framework designed for cluster-based computing environments in cloud data centers. It is specifically targeted for graph data analysis problems. The Beehive framework provides the abstraction of key-value based global object storage, which is maintained in memory of the cluster nodes. Its computation model is based on optimistic concurrency control in executing concurrent tasks as atomic transactions for harnessing amorphous parallelism in graph analysis problems. We describe here the architecture and the programming abstractions provided by this framework, and present the performance of the Beehive framework for several graph problems such as maximum flow, minimum weight spanning tree, graph coloring, and the PageRank algorithm.
ISSN:0190-3918
DOI:10.1109/ICPPW.2014.50