Hybrid storage engine for geospatial data using NoSQL and SQL paradigms

Abstract

The design and implementation of services to handle geospatial data involves thinking about storage engine performance and optimization for the desired use. NoSQL and relational databases bring their own advantages; therefore, it is necessary to choose one of these options according to the requirements of the solution. These requirements can change, or some operations may be performed in a more efficient way on another database engine, so using just one engine means being tied to its features and work model. This paper presents a hybrid approach (NoSQL-SQL) to store geospatial data on MongoDB, which are replicated and mapped on a PostgreSQL database, using an open source tool called ToroDB Stampede; solutions then can take advantage from either NoSQL or SQL features, to satisfy most of the requirements associated to the storage engine performance. A descriptive analysis to explain the workflow of the replication and synchronization in both engines precedes the quantitative analysis by which it was possible to determine that a normal database in PostgreSQL has a shorter response time than to perform the query in PostgreSQL with the hybrid database. In addition, the type of geometry increases the update response time of a materialized view.

Evaluation of the response time of a geoservice using a hybrid and distributed database

“…The implementation of the hybrid and distributed database has a sense of synchronization from MongoDB to PostgreSQL, taking advantage of the ToroDB tool that performs a decomposition process from documents to relational tables. To unify the data in a single table, triggers were implemented to keep the data synchronized in materialized views that are consumed by the GeoToroTur geoservice (Herrera-Ramírez et al, 2021).…”

Cloud based evaluation of databases for stock market data

“…Currently, efforts are being made to merge the two database systems to offer the best of both worlds [45,92], where, for example, a hybrid model would provide the flexibility that is prevented by the rigid relational database framework [54]. Most recently, a hybrid database was implemented where simple requests (read, insert) were served by MongoDB, while complex operations, such as joins with filtering the requests, were forwarded to PostgreSQL [43]. These hybrid models integrate SQL and NoSQL databases in one system to eliminate the limitations of individual systems.…”