Cosmos DB: Cost Optimization Strategies

Cosmos DB can be expensive if not configured correctly. Here’s how to optimize costs while maintaining performance.

Understand RU/s Consumption

• Monitor consumed vs provisioned RU/s
• Use Azure Monitor metrics
• Enable per-partition metrics

Right-Size Throughput

• Autoscale: For variable workloads (10x range)
• Manual: For predictable workloads
• Serverless: For dev/test or sporadic usage

Query Optimization

// Check RU charge
var response = await container.ReadItemAsync<Item>(id, pk);
Console.WriteLine($"RU Charge: {response.RequestCharge}");

// Reduce RU with projections
var query = container.GetItemQueryIterator<Item>(
    new QueryDefinition("SELECT c.id, c.name FROM c WHERE c.category = @cat")
        .WithParameter("@cat", category));
// Smaller response = fewer RUs

Other Tips

• Use TTL for automatic data expiration
• Composite indexes for complex queries
• Consider reserved capacity for predictable workloads (65% savings)

References

• Optimize Cosmos DB Costs