Azure Cosmos DB provides guarantees for performance and latency, with upper bounds on operation. This guarantee is possible when the engine can enforce governance on the tenant's operations. This is possible based on setting the throughput, which ensures that you GET the guaranteed throughput and latency, because platform reserves this capacity and guarantees operation success. When you exceed this capacity you get overloaded error message indicating your capacity was exceeded. 0x1001 Overloaded: the request cannot be processed because "Request Rate is large". At this juncture it is essential to see what operations and their volume causes this issue. You can get an idea about consumed capacity exceeding the provisioned capacity with METRICS on the portal. Then you need to ensure capacity is consumed nearly equally across all underlying partitions. If you see most of the throughput is consumed by one partition, you have SKEW of workload.

Metrics are available that SHOW you how throughput is USED over hours, days, and per seven days, across partitions or in aggregate.

Azure Cosmos DB provides guarantees for performance and latency, with upper bounds on operation. This guarantee is possible when the engine can enforce governance on the tenant's operations. This is possible based on setting the throughput, which ensures that you get the guaranteed throughput and latency, because platform reserves this capacity and guarantees operation success. When you exceed this capacity you get overloaded error message indicating your capacity was exceeded. 0x1001 Overloaded: the request cannot be processed because "Request Rate is large". At this juncture it is essential to see what operations and their volume causes this issue. You can get an idea about consumed capacity exceeding the provisioned capacity with metrics on the portal. Then you need to ensure capacity is consumed nearly equally across all underlying partitions. If you see most of the throughput is consumed by one partition, you have skew of workload.

Metrics are available that show you how throughput is used over hours, days, and per seven days, across partitions or in aggregate.