Optimization of the geoinformation service in the real estate restoration support system

This article is a continuation of a series of studies aimed at developing a support system for the restoration process of real estate objects. The focus of the research is on the geoinformation service of this system. Approaches to optimizing the geoinformation service are examined, taking into account the specifics of time series analysis and resource metrics. It is demonstrated that the unique characteristics of the geoinformation service and the selected metrics necessitate an individualized approach to scaling and optimization, thereby ensuring high performance, adaptability, and cost-effectiveness of the system. Key parameters are identified whose comprehensive analysis facilitates efficient resource management, enhanced performance, and a reduction in the risk of execution delays. An integration of statistical forecasting methods, such as exponential smoothing and autoregressive integrated moving average, is proposed for predicting peak load periods based on data concerning memory usage, query processing duration, and the frequency of satellite image acquisitions. The application of exponential smoothing is shown to enable rapid responses to changes in the periodicity of image acquisitions and to forecast subsequent intervals with sufficient accuracy. Moreover, it is demonstrated that representing time series data in the form of Japanese candlesticks offers several advantages over traditional graphs. Such a representation of the dynamics of "memory usage" and "satellite image acquisition time" provides comprehensive insights into resource consumption trends. The practical significance of this work lies in ensuring continuous resource monitoring of the service, adaptive management, and the prompt updating of real estate object information. In this context, adaptive resource management combines proactive forecasting with reactive scaling to reduce delays and enhance overall system efficiency.