应用

基准程序

资源特征

Micro benchmarks

Sort

Intensive I/O, middle CPU

Word count

Intensive CPU, networkand light I/O

Tera sort

Map and Shufflfle stage is intensive CPU

Web search

Nutch indexing

Map stage is intensive CPU

Reduce stage is intensive I/O

Page rank

Intensive CPU

Machine learning

Bayesian

Intensive I/O

Classifification

The first job is intensive CPU

K-means clustering

Center point intensive CPU

Clustering is intensive I/O

HDFS

Enhanced DFSIO

Intensive I/O