Evaluating MapReduce for Multi-core and Multiprocessor...
 |
||
Google Video tech TalksFebruary 27, 2007ABSTRACTThis paper evaluates the suitability of the MapReduce model for multi-core and multi-processor systems. MapReduce was created by Google for application development on data-centers with thousands of servers. It allows programmers to write functional-style code that is automatically parallelized and scheduled in a distributed system.We describe Phoenix, an implementation of MapReduce for shared-memory systems that includes a programming API and an efficient runtime system. The Phoenix runtime automatically manages thread creation, dynamic task scheduling, data partitioning, and fault tolerance across processor nodes. We study Phoenix with...
Canal: Howto & Style
Añadido: November 30, 1999 at 12:00 am
Autor: googletechtalks
Duración: 44:47
Puntuación: 5.00
Reproducciones: 2225
Etiquetas: core evaluating google howto mapreduce multi
Comentarios
| No comments. |