How one team cut Azure Cosmos DB costs by 60%

Name: How one team cut Azure Cosmos DB costs by 60%
Uploaded: 2026-05-14T20:25:36+00:00
Description: Patty Chow recaps an Azure Cosmos DB Conf story where a team reduced Cosmos DB costs by 60% while eliminating throttling and dramatically improving P99...

Today by Patty Chow

Patty Chow shares a quick Cosmos DB Conf recap showing how an app can look healthy while still wasting money, and how one team fixed it through design changes rather than scaling.

Overview

A team improved both cost and performance in Azure Cosmos DB:

Cost reduction: 60% lower spend
Throttling: reduced from 142 errors to 0
Latency: improved P99 from 284 ms to ~16 ms

Key areas highlighted for Cosmos DB optimization:

RU/s tuning (align throughput with actual workload)
Partition key design (avoid hot partitions and uneven distribution)
Indexing policy (index what you query; avoid unnecessary indexing work)
Query patterns (shape queries to match partitioning and indexing)
Serverless vs. provisioned throughput trade-offs