An Electric Power Big Data Deployment Solution for Distributed Memory Computing

In the Big Data computing, improving performance with memory computing is one of hot spots. In the memory computing, the data deployment directly affects load balance and task efficiency. In the scene of memory computing of electric power data, two unsolved problems are: (1) only memory space, witho...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Proceedings (International Symposium on Parallel Architectures, Algorithms and Programming. Print) s. 155 - 161
Hlavní autoři: Zhi Yang, Chunping Zhang, Mu Hu, Feng Lin
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.12.2015
Témata:
Big data
Business
Data Deployment strategy
Distributed databases
Distributed Memory Computing
Hardware
Objectification Parallel Computing
Parallel processing
Power systems
ISSN:2168-3042
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In the Big Data computing, improving performance with memory computing is one of hot spots. In the memory computing, the data deployment directly affects load balance and task efficiency. In the scene of memory computing of electric power data, two unsolved problems are: (1) only memory space, without the CPU frequency and nuclear number, could be considered for load balance and improving performance, (2) there are so many manual operations that it is difficult to complete data deployment automatically. This paper provides an electric power data deployment solution for distributed memory computing to solve the above challenges. In the solution, according to business logic and hardware configuration of cluster nodes, the data deployment strategy can be established. Then, the deployment scheme can be implemented with interface operation. Lastly, cluster nodes load data according to the deployment scheme. The solution has been applied to the Objectification Parallel Computing (OPC). The application result shows that OPC can achieve the best performance which can meet the demand of system efficiency and the operation of data deployment is simple.
ISSN:2168-3042
DOI:10.1109/PAAP.2015.38