Fig. 6
From: Upgrading a high performance computing environment for massive data processing
Speedup as a function of the workload size for Grep and Wordcount applications, considering Python and Java implementations, when the HDFS API is configured with (S) or without (H) the Storage API, and with no block replication (1) or 3-way replication (3). Speedup is computed compared to the execution times of the same applications executing on a conventional file system. a Grep - Python. b Grep - Java. c Wordcount - Python. d Wordcount - Java