The growth of digital transformation in e-government implies rapid evolution of IT systems. In order to ensure the secure operation and high availability of these systems require the integration of Machine Learning (ML) techniques and AIOps for reducing the operation cost, improving the reliability and availability of the services. Our initiative is to develop, improve, and integrate new or existing AI-based methods using open source technologies for the analysis of machine-generated telemetry data (logs, metrics, and traces), which is one of the main pillars of protecting the national data assets for fulfilling the following requirements:
- Faster, more efficient deep learning-based error handling, faster response times for
- solving problems, and shorter downtime of the services.
- Unified data analysis and visualization
- Reducing the cost of business losses due to unexpected intervention
- Additional costs avoided by timely intervention
- Regular feedback into the development process, improving the quality of the
- delivered software
- Continuous feedback on system operation, and automatic generation of weekly reports maximize the pro-activity
Taking into account these requirements, we have designed the MaLOM system with an elastic architecture that allows data collection, transformation, and storage in a big data platform and supports combined data analysis capabilities. MaLOM focuses on using Deep Learning (DL) through a web interface and exploits the training of neural networks using MLOps. Using well-defined protocols, in addition, we have also introduced an alerting system for notifying the appropriate person in case of abnormal events.
