Skip to main content
Each region you add to an Azure Cosmos DB account is billed again for its provisioned throughput (RU/s) and stored data. nOps finds accounts billed in two or more regions every day and estimates the saving from removing every region except the one that handles writes. Extra regions are usually there for read latency or disaster recovery, so this is most often worth doing for non-production accounts.

Why it matters

A second region roughly doubles an account’s throughput and storage cost, and it is easy to leave in place on development and test accounts that were copied from production.

How nOps finds it

nOps reads the last 7 days of pay-as-you-go Cosmos DB billing. An account is flagged when its throughput was billed in at least two regions on at least 6 of those days. The kept region is the one that costs the most (on a near tie, the region billing shows as the write region). The saving is the throughput and storage cost of the other regions. Throughput covered by reserved capacity is not counted.

Before you start

Confirm with the account owner whether the extra regions are needed for read latency or disaster recovery.

How to implement

  1. Confirm no application or recovery plan relies on the extra regions.
  2. If the account uses multi-region writes, turn them off first.
  3. In the Azure portal, open the Cosmos DB account and go to Replicate data globally.
  4. Remove the extra regions, keeping the write region, and save.

Risks and considerations

  • Removing a region removes its local reads and its role in failover.
  • You can add a region back later. Cosmos DB copies the data to it again, which takes time and costs data transfer.

FAQs

The region that costs the most, which is normally the write region. On a near tie, the region billing shows sending replication traffic.