MR2 Job Trace - Amithsha/MR2-Jobs-Traces GitHub Wiki

###Description For the cluster performance comparison instead of creating the random no's of jobs or using the default benchmarks jars, old concept called run the jobs using the job traces in MR1 is renewed by creating a python script to trace the MR2 Jobs.

yes!! we can run the jobs as same in production. with n no of times to do benchmark

###SWIM Statistical Workload Injector for MapReduce This SWIM supports for MR1 It allows realistic and representative workloads to be used to drive such performance comparisons. Please refer the link for SWIM and how to use https://github.com/SWIMProjectUCB/SWIM/wiki.

###Bird's Eye view of SWIM

1. Pull the job Trace
2. Convert to Samples
3. Create data for samples + Create script to Run the samples

Here in the 1st step pulling the job trace is MR1 not available for MR2 so created a python script which will replace the swim perl script and produce the output in tsv