Build I/O-Aware Analytic Model for Apache Spark Applications, and we show GATK4 (Apache Spark based GATK) developed by Broad Institute as the driving example.
- I/O profiling (we provide I/O profiling result for Google Cloud)
- Per-Application profiling Sampling runs are used to charaterize the performance, after training on analytical model, predict the performance and assocaited cost.