Change in Azure SQL Database serverless auto-pause resume latency when adjusting delay

Goal: Choose an auto‑pause delay for the Azure SQL Database serverless tier that minimizes compute cost for a sporadic workload without introducing unacceptable resume latency for end‑users.

Uncertainty: The time required to resume a paused database after the first request is not fixed; observations show a range from a few seconds to more than a minute, influenced by the underlying compute hardware and the Azure region where the database resides. This variability makes it difficult to predict whether a given delay setting will keep latency within service‑level expectations.

Questions: What is the maximum resume latency that the application can tolerate? How does resume latency differ across regions and hardware generations for the same auto‑pause delay? Does increasing the auto‑pause delay beyond the default 60 minutes reduce the frequency of resume events enough to offset the latency penalty?