Využití bezserverových funkcí pro zpracování geografických dat

Abstract

The aim of this bachelor's thesis is to test the functionality and explore the practical possibilities of using serverless architecture (the FaaS - Function as a Service computing model) for spatial data processing. To facilitate this testing, a web geoinformation application based on a serverless cloud architecture was designed and implemented, which allows asynchronous spatial data processing and dynamic rendering of satellite images. The thesis first explores cloud computing technologies and the FaaS (Function as a Service) model. Subsequently, the practical implementation of the solution using Amazon Web Services (AWS Lambda, Amazon S3), a Django web backend, and the Leaflet interactive map interface is described in detail. Vector geoprocessing in the cloud includes the asynchronous creation of buffer zones with dynamic reprojection, while the raster part handles the on-the-fly calculation of the NDVI vegetation index from gigabyte-sized Sentinel-2 satellite images using the Cloud Optimized GeoTIFF (COG) format and Docker containerization. The results of the stress tests provide practical verification of the extreme scalability, zero maintenance, and cost-efficiency of the serverless approach compared to traditional GIS servers.