Распрацоўка шматкампанентнай ГІС з маштабаваннем на аснове віртуалізацыі

Abstract

A distributed architecture for dynamic run of the geoinformation system elements based on the Kubernetes container virtualization platform is presented. The system architecture was developed within the flood monitoring and prediction system. The system core includes a database cluster with high availability based on the multi-server PostgreSQL relational database management system with asynchronous replication and the PostGIS extension to store geospatial data. The interaction of system components is discussed as well as the integration of pluggable modules to import data streams from external sources, and computing modules for updating the geodata layers based on the flood zone calculations. The issues of horizontal scaling and automation of planned system upgrades are discussed, as well as the possibility of integrating specialized time series storage tools.