Provisioned throughput is a fixed amount of request-processing capacity, measured in request units per second (RU/s), allocated to an Azure Cosmos DB container or database. It fits workloads with predictable, consistent traffic because capacity is chosen in advance. section 11.6.1
Choosing provisioned throughput means accepting a little planning in exchange for predictable capacity:
- Allocate enough RU/s for expected demand without over-provisioning. section 11.6.1
- Prefer it for stable traffic patterns, where fixed capacity can be cost effective. section 11.6.1
- Reconsider it when demand is unpredictable: autoscale can adjust throughput up to a defined maximum and reduce manual intervention during spikes. section 11.6.1
- Watch costs carefully, because higher request-unit consumption during busy periods can increase the bill. section 11.1.3
C:/> az cosmosdb sql database throughput update --account-name alpacacosmosdbaccount --name alpacacatalogdb --resource-group AlpacaRG --throughput 1000
The Azure CLI updates database throughput with --throughput, making the configured capacity explicit; the paired autoscale form uses --max-throughput instead. section 11.6.1
A steady product catalog
Alpaca-Ma-Bags allocates 1,000 RU/s to its product catalog database during off-peak times. The capacity covers expected demand without paying for unnecessary throughput; holiday traffic may call for autoscale instead. section 11.6.1
