Benchmarking MongoDB vs ScyllaDB: IoT Sensor Workload Deep Dive

BenchANT's benchmarking analysis highlights the performance and scalability differences between MongoDB and ScyllaDB for an IoT sensor workload, with ScyllaDB generally outperforming MongoDB in terms of throughput and latency. The study utilized YCSB's default data model with a focus on 90% insert operations and 10% read operations, simulating real-world IoT applications. ScyllaDB demonstrated significantly higher throughput, achieving up to 19 times that of MongoDB and lower update latency, while MongoDB offered lower read latency in smaller scaling sizes but was surpassed by ScyllaDB at medium scaling. Cost analysis revealed that ScyllaDB provides more operations per dollar compared to MongoDB, especially as the data size increases. The benchmark also explored the impact of data models, suggesting that a pure key-value approach offers modest throughput improvements for ScyllaDB in write-heavy workloads, whereas MongoDB showed no significant benefit. Overall, ScyllaDB's performance suggests superior scalability and cost-efficiency for IoT applications compared to MongoDB.