<![CDATA[Dynamic form builders require database structures that can accommodate variable attributes while maintaining efficient query performance. However, the performance implications of different storage representations within the same database management system remain insufficiently characterized, particularly for relational, raw JSON, and hybrid relational-JSON models. This study evaluates the performance of these three storage representations for dynamic-form data using MariaDB 10.6.19. A controlled quantitative experiment was conducted using datasets of 10,000, 50,000, and 100,000 records and six read and aggregation workloads covering point retrieval, dynamic-field search, multi-field filtering, counting, averaging, and grouping. Each workload was executed using 10 warm-up iterations followed by 30 measured repetitions. Query latency, storage consumption, execution plans, and statistical differences among representations were analyzed. The results show that performance varied according to workload and dataset scale. Statistically significant differences were observed in 13 of 18 workload-dataset combinations. The representation with the lowest median latency also varied across conditions, indicating that no single storage representation consistently provided the lowest query latency. Execution-plan analysis for DYNAMIC_FIELD_SEARCH at 100,000 records showed indexed access for relational storage and full-table scans for the evaluated raw JSON and hybrid configurations. Raw JSON required the lowest storage consumption, followed by the hybrid representation. These findings indicate that storage representation should be selected according to workload characteristics, dataset scale, and structural requirements rather than based on a single overall performance criterion.]]>
Copyrights © 2026
