This project evaluates the performance of three managed NoSQL databases—Amazon DynamoDB, MongoDB Atlas, and Apache Cassandra (Astra DB)—using an e-commerce dataset of ~51,000 records. We performed empirical testing to determine how their unique architectures impact latency and throughput across CRUD and bulk operations.
No single database dominated all categories, confirming that technology selection depends on specific workload requirements:
Amazon DynamoDB: Fastest for single-item writes (291.91 ms), range scans (376.97 ms), and deletions (38.37 ms).
MongoDB Atlas: The "champion" for bulk data loading (939.11 ms), handling high-volume ingestion most efficiently.
Astra DB (Cassandra): Demonstrated the best performance for specific ID lookups (382.06 ms).
DynamoDB is ideal for real-time transactional hotspots like carts and sessions.
MongoDB Atlas excels in flexible querying and complex analytical workloads.
Astra DB is preferred for write-intensive, globally distributed scenarios such as IoT telemetry.