Scalable Data Management in Dataspaces: Benchmarking MongoDB Sharding

Edge-powered dataspaces enhance time-sensitive applications, efficient bandwidth, security, and privacy, particularly for smart urban environment data spaces. In addition, designing optimized distributed edge-powered dataspaces that consider the shared data residing near the provider’s ease the management and control over the data. However, since distributed edge devices are heterogeneous, application deployment and management are typically challenging. Container virtualization is an important technology that addresses this problem, specifically on devices with limited resources. Scaling the number of nodes does not always guarantee adequate performance improvements. Therefore, it is crucial to study the effect of horizontal scalability on runtime, throughput, latency, and scaling the record and operation counts. In this study, the MongoDB database has been benchmarked using Yahoo! Cloud Serving Benchmark with a Docker Container of a single node and a Docker Swarm cluster of nine nodes scaled horizontally on Raspberry Pi. In addition, the study scales the record and operation counts to understand their effect in a similar environment.