English | EN 
Cloud computing offers a ubiquitous, convenient, and on-demand network model, enabling organizations to access necessary services more efficiently and eliminate the need for costly and space-consuming on-site hardware. Consequently, in recent years, cloud computing has been widely adopted across various industries and has driven a significant evolution in how organizations access, manage, and develop their information resources. However, managing cloud spending, budgeting and usage has become increasingly challenging for organizations due to the complexity of cloud cost models, the diverse range of cloud services, and the decentralized decision-making processes regarding cloud usage within organizations. Advances in Artificial Intelligence (AI) and Explainable Artificial Intelligence (XAI) propose promising solutions in terms of efficiency and trustworthiness.
How can AI and XAI enable optimal cloud usage?
Cloud Cost Forecasting
The application of AI in cloud cost forecasting remains relatively novel but promising, as AI models can analyze large volumes of historical and real-time data to identify consumption trends and variation across organizations. Additionally, AI models can incorporate factors such as seasonal changes, market conditions, and shifts in resource usage to generate accurate, adaptive cost predictions that align with the evolving cloud environment. When developing an AI-driven forecasting method to predict future cloud costs, ensuring the trustworthiness and transparency of predictions is essential to comply with social and ethical standards in the finance sector. Therefore, integrating black-box AI models with Explainable AI (XAI) methods to provide interpretable insights into predictions becomes crucial.
Evaluating the effectiveness of current usage
AI models can analyze historical usage metrics, including consumed CPU percentages and storage capacity, to understand usage patterns and behaviors. This analytical process yields valuable insights into the effectiveness of cloud resource utilization in organizations. The evaluation goes beyond comparing consumed versus reserved capacity; it also considers usage evolution to avoid over- or under-utilization. Furthermore, the integration of workload forecasting into this evaluative framework shows promise in enabling appropriate recommendations for resource adjustments.
Recommendations for resource adapted to requirements
The cloud offers a wide variety of resources with diverse characteristics in terms of performance, security, and compliance, making the selection and estimation of resources particularly complex. Therefore, identifying the appropriate resources that match the needs is essential to optimize expenses and avoid costly consumption. An AI-based recommendation model can learn from cloud providers’ documentation, analyze the needs, and make recommendations about the most suitable resources. This solution can lower costs and reduce resource wastage.
Anomaly Detection
Anomalies in cloud usage can arise from abnormal participant behaviors. An AI-based anomaly detection model aims to identify normal behaviors in historical data, analyze current costs and usage over a period, and compare them with established normal behaviors to determine if an anomaly exists. This method can provide instantaneous alerts to organizations, empowering them to take quick action in case of an anomaly. Consequently, it allows for adjustments in usage behaviors to align with project strategy and budget.
Conclusion
Integrating advanced AI techniques with XAI methods offer a wide range of promising solutions with accurate and reliable support to help organizations understand their usage behaviors, optimize their consumption to reduce costs, and select the most appropriate resources for their needs. Furthermore, by using cloud services efficiently, organizations can contribute to reducing consumed energy, empowering green