Performance Evaluation of NoSQL Databases as a Service: Couchbase Capella and MongoDB Atlas

This technical report compares the throughput and latency of two NoSQL databases as a service (DBaaS) across four different workloads and three cluster configurations.

To get the full document

or fill out the form

Why read this?

NoSQL systems were created to support the demands of modern, globally distributed, cloud-native applications. While NoSQL solutions help to maximize performance and uptime for modern application workloads, database-as-a-service alternatives may assist in saving time and money on cluster support, deployment, and maintenance.

This study evaluates the latency and throughput of Couchbase Capella™ (March 2022 release) and MongoDB™ Atlas v5.0 under four workloads:

an update-heavy workload that simulates scenarios for a typical e-commerce application
a short-range scan workload that simulates threaded conversations
a pagination workload that simulates a selection by field with pagination
a JOIN workload with grouping and ordering that simulates a selection of complex child–parent relationships with categorization

The comparison was conducted on three cluster configurations: 6, 9, and 18 nodes. For consistency, the Yahoo! Cloud Serving Benchmark was utilized. The performance results are supported by seven descriptive tables and four comparative diagrams.