Computação intensiva em dados com MapReduce em ambientes oportunistas

Jonhnny Weslley SilvaThiago Emmanuel PereiraCarla de Araújo SouzaFrancisco Brasileiro

The MapReduce programming model has emerged as a popular approachfor the processing and generation of large data sets. Considering the opportunisticscenario, the execution of such data intensive application can harmuser experience and impact application's makespan as a consequence of naivedata placement and scheduling. To deal with these problems, we have proposedand analysed data allocation strategies which consider resource volatility. Wealso have modified and evaluated BashReduce, a light weight implementation ofMapReduce model based on UNIX bash tools. Our modified version had its performanceevaluated against Hadoop showing a 36% speedup on application'smakespan.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: