linkedin / dynamometer

A tool for scale and performance testing of HDFS with a specific focus on the NameNode.
BSD 2-Clause "Simplified" License
131 stars 34 forks source link

Create a new TimedInputFormat with the time-based virtual split logic used by CreateFileMapper. This allows for future workloads to easily also have time-based mapper logic. #75

Closed xkrogen closed 5 years ago

xkrogen commented 5 years ago

The CreateFileMapper is one example of a workload in which we want X number of mappers to do some action for X minutes, but this is a generally useful workload pattern. Pull out the time-based logic so that future workloads can simply define the type of action to take, and leverage the time-based behavior of CreateFileMapper.