Many clusters offer the ability to scale up and down job containers. If Dynamic Allocation is turned on you may not need or desire Owl's recommended
executor-memory. However, in our testing right sizing the job before executing is both faster and a healthy habit. Faster, because there is less orchestration and context switching while the job is doing work; we've minimized time spent running out of space and having to reallocate and shuffle to a new container. Healthier, because it give the user real-time feedback on the cost of each feature and the ability to control the cost benefit analysis.