Optimized Spark configuration to enable Data Science at very high performance

Description

Changing the default configuration for Spark 2.x to drive higher performance

Use Case and Business Need

As Spark changes occur and improvements are made, and more and more workloads start using Spark as the engine, can Oracle - continuously - optimize the configurations to enable higher performance for workloads that rely on Spark? This enables customers to not spend time on tuning, but to simply see better performance after upgrades of their systems. This has a big impact on data science workloads - both model building and model execution - and on more mundane tasks around data transformation etc.