In the following repositories you will find the components created for anaylizing, controlling and evaluate our tool.
- xSpark: an extension of Spark for controlling big-data applications
- xSpark-bench: a tool to run benchmarks and evaluate xSpark
- Spark-seq: a simple Spark fork for the sequential execution of Spark stages
- xSpark-dock: containerized Spark executors
- Spark-perf: spark-perf benchmark with the ability of generating dataset with skewed distribution (Zip'f law)
- Dilworth-labeling: DAG labeling with Dilworth's theorem