An Azure App Service intermittently failed to establish a TDS connection to Microsoft Fabric SQL within 15 seconds. Review Queue and Worklists also returned HTTP 500 errors after approximately 15 seconds. This occurred before any SQL statements executed and was not related to login or permissions. The issue stemmed from Fabric platform throttling due to heavy SQL workloads on a small, shared capacity. Microsoft Fabric provides compute through bursting, smoothing, and carry-forward usage. When smoothed usage exceeds capacity, operations are delayed or rejected. To mitigate active throttling, users can wait for capacity to self-heal, temporarily increase the F SKU size, or pause and resume the capacity. Capacity overage billing is also an option. To improve application resilience, implement bounded retries with jitter, align SQL timeouts with application budgets, and capture detailed logs. Monitoring capacity utilization, throttling events, and overages is crucial using the Microsoft Fabric Capacity Metrics app. Users should configure capacity alerts for when provisioned CU usage reaches 100%. Reducing peak concurrent usage by staggering operations and distributing workloads can help prevent throttling.