Autoscale throughput

  • 2 minutes

We can often make an educated guess about where our workload is in terms of throughput, but we don’t know exactly where it lands until it’s in production. We also may know our operational tolerances. We know:

  • The maximum amount of money we're willing to spend
  • The minimum amount of performance we're ready to tolerate.

With all of this information in mind, we could define a range. This range would represent our application running at a comfortable performance level without overspending without our knowledge.

Application with usage oscillating between maximum potential spend and minimum performance

With Azure Cosmos DB autoscale, we can define a range of request units per second (RU/s) to scale our container automatically and instantly. The throughput RU/s is scaled based on real-time usage instantly.

Autoscale between max and min RU/s

Autoscale is great for workloads with variable or unpredictable traffic patterns and can minimize unused capacity that would typically be pre-provisioned.