Skip to main content

Advertisement

Table 2 Cloud emulating Hadoop benchmarks: I/O characteristics

From: Host managed contention avoidance storage solutions for Big Data

Workload I/O characteristics
Grep Mostly sequential reads with small writes
Random text writer Mostly sequential writes, mixed with random writes and negligible reads
Sort More reads than writes. Large sequential reads with random writes and later sequential writes
TeraSort Good mix of sequential and random reads/writes. More reads than writes
Wordcount Mostly sequential reads, with large number of random writes followed by random reads and small sequential writes
Word standard deviation Mostly sequential reads with small inter-phase writes, followed by small writes in the end