School-district attendance ETL must burst during enrollment week without unbounded spend. How should worker counts be set?

School-district attendance ETL that must burst during enrollment week without unbounded spend should configure min and max workers so the job can scale within a cost-controlled range. Minimum workers cover baseline throughput; maximum workers cap enrollment-week scale-out so the district can absorb file spikes without an uncapped cloud bill—the dual requirement in the stem.

Setting minimum workers for the enrollment baseline keeps the cluster sized for enrollment week even when attendance volume is low, so it spends at enrollment-week levels year-round, and usage budget alerts only notify after spend instead of enforcing a max-worker ceiling for cost control. Fixing minimum and maximum workers to the enrollment-week forecast locks the cluster at a single forecasted size, so it cannot scale down when attendance volume is low and cannot burst beyond forecast when files exceed predictions, losing the elastic cost-controlled range. Enabling autoscaling with a large step and short cooldown tunes only how quickly workers are added, without setting the minimum/maximum worker guardrails, so aggressive scale-out can overshoot cost and thrash while still lacking a cost-controlled burst range.

Exam caveat: autoscaling with a large step and short cooldown tunes speed without setting bounds when burst within cost control is in the stem; fixing min and max to the enrollment-week forecast is static and wasteful when elastic range is in the stem. Correct answers usually cite both min and max. SQL warehouse scaling has its own knobs—here the stem is worker counts on ETL/job cluster compute.

Operational check: set attendance job-cluster min/max from observed enrollment-week peaks plus budget, enable autoscaling, alert when max workers are saturated, and revisit bounds after enrollment week using job metrics and cost exports.