Geospatial Data Exchange Using Binary Data Serialization Approaches

Abstract

Abstract. In this paper we investigate the benefits of binary data serialization as a means of storing and sharing large amounts of geospatial data in an interoperable way. De-facto text-based exchange encodings typically exposed by modern Application Programming Interfaces (APIs), including eXtensible Markup Language (XML) and JavaScript Object Notation (JSON), are generally inefficient for an increasingly higher number of applications due to their inflated volumes of data, low speed and the high computational cost for parsing and processing. In this work we consider comparisons of JSON/Geospatial JSON (GeoJSON) and two popular binary data encodings (Protocol Buffers and Apache Avro) for storing and sharing geospatial data. Using a number of experiments, we illustrate the advantages and disadvantages of both approaches for common workflows that make use of geospatial data encodings such as GeoPackage and GeoJSON. The paper contributes a number of practical recommendations around the potential for binary data serialization for interoperable (geospatial) data storage and sharing in the future.

Lite2: A Schemaless Zero-Copy Serialization Format

“…Firstly, Apache Avro is a language-neutral data serialization system. Thanks to its efficient binary format, it does not require code generation and can interoperate with various programming languages [27]. Secondly, it has better performance in data file size and reading/writing execution time when compared with other data formats such as XStream [28], Protostuff [29], Thrift, and so on [30].…”

“…However, Apache Avro also has a significant drawback-heavy dependence on the schema. Schema designing and updating are of high importance in the application of Apache Avro, and the binary data implementation needs specified knowledge and experience [27].…”

Interactive Geoinformation System for Dynamic Visualization of Auroral Oval Characteristics Based on Component-Oriented Programming Patterns

“…The initial data are the results of satellite observations, available via standard network data transfer protocols and provided by providers in accordance with a given time stamp in the form of a set of attribute and spatial information. The data provided by the services is formatted in text data streams in CSV-like and/or JSON format [12], which allows them to be efficiently processed in appropriate server scripts.…”

“…The model itself is implemented at the level of software scripts executed on the server side and returns a text dataset in GeoJSON format as a result [12]. Each geospatial primitive (spatial point) in the corresponding server response represents a combination of three components -geographic latitude and longitude (spatial data) and the probability value of observing the aurora at the corresponding point in space.…”