Tearing Down the Tower of Babel: Unified and Efficient Spatio-temporal Queries for NoSQL Stores

NoSQL stores are used extensively for scalable storage and efficient querying of large spatio-temporal data collections in modern applications. Yet, despite their popularity, NoSQL systems have two main limitations when confronted with spatio-temporal data: (a) they do not offer optimized indexing methods, and (b) they still rely on heterogeneous languages and lack of standardization in data access, a situation bearing resemblance to the narrative of the tower of Babel. To address these limitations, we propose NoDA, a system for scalable querying of spatio-temporal data stored in different NoSQL stores in a unified way. NoDA relies on an abstraction layer that consists of data access operators with clear semantics, that provides a unified view of the underlying NoSQL stores. Furthermore, NoDA offers spatio-temporal operators that are internally implemented in an efficient way, by taking into advantage the individual features of each NoSQL store. Capitalizing on the query operators, NoDA provides a declarative interface based on a SQL-like language, allowing users to query different NoSQL stores using SQL. Our experiments demonstrate that NoDA significantly improves the performance of spatio-temporal querying over different types of NoSQL stores.