A PERFORMANCE EVALUATION OF TWO EQUI-JOIN MODELS FOR PROCESSING BIG DATA

This repository contains two solutions for implementing an equi-join for processing big data. Both solutions were written in C++. The Hybrid solution makes use of MPI and OpenMP. The MapReduce Solution uses Phoenix++ to implement the solution. The peformance of the two solutions were benchmarked and compared in the Group 11 ELEN4020 Project Report pdf document.

Comparison of the two solutions over various file sizes

Branches:

Documentation: Contains the .tex and .pdf files of our report for our project.

Master: Contains content of merged branches in relevant folders:

Hybrid Solution: Contains the code for the MPI-OpenMP Hybrid solution.

MapReduce Solution: Contains the code and the required libraries and header files for the MapReduce solution.

Documentation: All documentation related files, including the project report

Test Input: Contains two test files for running the solutions

Compilation and Execution Instructions:

Hybrid Solution:

To compile:

make mpi-proj

To run:

mpirun --hostfile [name of hostfile] -np [number of processes] ./Project [input file 1] [input file 2] [column containing key]

OR

mpiexec -n [number of processes] -f [name of hostfile] ./Project [input file 1] [input file 2] [column containing key]

MapReduce Solution:

To compile and run:

./MapReduce.sh [input file 1] [input file 2] [column containing key]