Design and Implementation of a Distributed Database Benchmark

With the complexity and change of business scenarios, compared to traditional databases that can only process a single type of data, multimodal databases that can store and process multiple types of data at the same time are increasingly favored by various business platforms. Therefore, it is very important to build a multimodal database benchmark test for evaluation, and the multimodal benchmark test in recent years has higher performance requirements for a single-machine environment. Therefore, this paper proposes a distributed architecture database benchmark to reduce the requirements for single-machine performance while making reasonable evaluations of multimodal databases. This article uses the social business scenario as a simulation scenario to build a multimodal data model that includes relational, graph, key-value, JSON and XML data and uses MapReduce to optimize the data generation algorithm. Finally, we conducted a data import test on a recent common multimodal database to show the usability of the benchmark.