The Monday 9am PileupWarehouse · medium · ~12 minEvery weekday 09:00-10:15, dashboard queries on BI_WH (size M, single cluster) queue badly — 200+ analysts log in at once. The rest of the day, the warehouse idles at 20% with zero queueing. The proposed fix on the table: 'permanently upsize to XL'. Snowflake rates: M = 4 cr/hr, XL = 16 cr/hr.Queues from concurrency need a different lever than queues from heaviness. Pick it, and price both options.Evidence

Metrics

Config & Rates

Checkpoints

Why is a permanent XL the wrong shape for this problem?

Price the sensible option: multi-cluster adds on average 2 extra M clusters (4 cr/hr each) for 1.25h/day. How many EXTRA credits per day is that? (2 x 4 x 1.25)

Compare: permanent XL adds 12 cr/hr over M for ~12 business hours ≈ 144 extra credits/day vs your multi-cluster figure. The senior recommendation?

What demand-side checks would you run before ANY capacity change? Interview-style.