What are the key components of the Spark execution model (Job, Stage, Task)?

Job: Triggered by one action (count, write, collect); one job per action. Stage: Boundary at shuffle; narrow transformations (map, filter) pipeline into one stage; wide (join, groupBy, distinct) trigger stage boundaries. Task: One task per partition per stage; unit of work on an executor. Flow: Action → Job → DAG Scheduler (plans stages) → Task Scheduler (schedules tasks to executors). Why it matters for tuning: Fewer stages = less overhead; more tasks = finer parallelism but more scheduling cost. Scalability: Stage count grows with shuffle count; task count = stages × partitions. Cost implication: Many small tasks (e.g., 10000 tasks of 1s each) waste time on scheduling; fewer large tasks (e.g., 200 tasks of 50s) are more efficient. Architectural logic: Minimize shuffles to reduce stages; tune partition count so tasks run 1–3 minutes. Best practice: Use Spark UI to inspect stage boundaries; identify shuffle-heavy stages for optimization.