zippond / db-engine-paradigms

Collection of experiments to carve out the differences between two types of relational query processing engines: Vectorizing (interpretation based) engines and compiling engines.

Database Engines: Vectorization vs. Compilation

This repository contains a collection of experiments, conducted to carve out the differences between two types of query processing engines: Vectorizing (interpretation based) engines and compiling engines.

Where to Start

Have a look at src/benchmarks/tpch/queries/ to see how query processing for Typer and Tectorwise works.

How to Build

A configuration file is provided to build this project with CMake. In the project directory run:

mkdir -p build/release
cd build/release
cmake -DCMAKE_BUILD_TYPE=Release ../..
make

This creates among others the main binaries test_all and run_tpch. Use test_all to run unit tests and check whether your compilation worked. Our main binary run_tpch requires TPC-H tables as generated by the TPC-H dbgen tool. With these our experimental queries can be run on arbitrary scale factors.