June, 1st - June, 7th

Writing of the Results chapter has begun while the collection of the results is continued. We have also tested the newly developed PcapInputFormat and have concluded that it behaves far better than the previous implementation. However, a strange behavior has been noticed with the rack local tasks, which seem to take far longer to execute (3 times more!) than data local.

After another debugging session, the Professor has been able to fix the issue. We have also added additional features to the Integrated Solution, such as the automatic retrieval of job logs and map task statistics, and generation of map task execution graphs organized by physical node. Currently the focus is to finish collecting results with Iperf and writing the sections relative to the Iperf traffic.

NEXT OBJECTIVES: After a meeting with the Professor, we have agreed to test the solution in heterogeneous clusters and compare the performance to homogeneous clusters. Furthermore, since the PcapInputFormat is working with real traffic, we will test our solution with real traffic and collect results as well.