Saeed Talatian Azad

In this study, the performance of four widely used databases Redis, PostgreSQL, Cassandra, and MongoDB is evaluated under six identical workloads. The primary objective is to provide a realistic, data-driven comparison of the throughput, latency, and resource consumption of these databases under strictly controlled conditions. All experiments were conducted on a dataset containing 500,000 records within an identical hardware and software environment to minimize the impact of external factors. The workloads included insert-only, read-only, and mixed read-update patterns, with each test repeated five times to record reliable averages and stable system behavior. Results indicate that Redis delivers the highest throughput and the lowest latency across all workloads, owing to its in-memory architecture. PostgreSQL demonstrates performance comparable to Redis in read-heavy workloads, albeit with higher CPU utilization. MongoDB exhibits balanced and stable behavior in terms of latency and resource consumption, while Cassandra shows steady performance in write-intensive workloads due to its LSM-Tree architecture. Overall, although Redis demonstrated a distinct advantage across all workloads in this study, the choice of a database should still be dictated by the workload pattern and operational requirements. The findings of this study can serve as a practical guide for system architects and developers in selecting the appropriate database engine.