Non-native Techniques for Storing JSON Documents into Relational Tables

The natural way, how JSON documents can be queried and modified is to store them first in relational environment. In such a case, the features of relational DBMSs such as transaction processing, can be used. In this paper we compare two different mapping techniques: Adjacency List and the Single-Table Data Mapping (STDM) algorithm, which can be used, among other techniques, to store JSON documents in relational tables. The reason to choose and compare these two techniques is due to their origin: both are representatives of two different non-native storing techniques. The former is a general technique, which can be applied to any data presented in hierarchical form, while the latter is a representative of the family of XML-to-Relational storage algorithms, which can be used for JSON, too. Our results show that using the STDM algorithm the size of resulting relational table is approximately 70% of the size of the corresponding table generated with Adjacency List. Additionally, the STDM algorithm significantly outperforms Adjacency List concerning time.