Multi-Cloud SLA-Based Broker Intelligently Translates Metrics, Overcoming Provider Lock-In for Cloud Consumers

Cloud computing underpins a vast majority of modern applications and services, yet realising its full potential remains challenging for many users. Víctor Rampérez, Javier Soriano, and David Lizcano, from Universidad Politécnica de Madrid and Madrid Open University, alongside Shadi Aljawarneh from Jordan University of Science and Technology and Juan A. Lara, address a key obstacle: the difficulty for cloud consumers to ensure consistent service levels across different providers.

The team presents a novel approach that automatically translates complex service-level agreements into measurable, vendor-neutral metrics, effectively removing the need for specialist expertise and preventing provider lock-in. This intelligent knowledge-based system not only provides a means of monitoring performance across multiple cloud platforms, but also offers feedback to users, acting as an intelligent tutoring system to optimise cloud resource allocation and unlock the benefits of multi-cloud environments, as validated through use cases involving leading cloud providers. Cloud SLAs, Auto-Scaling and Multi-Cloud Challenges This research comprehensively explores Service Level Agreements (SLAs), cloud computing, auto-scaling, and related technologies, revealing key themes and challenges in modern cloud environments. It focuses on the importance of clearly defined SLAs, examining how to translate high-level service objectives into measurable policies and metrics that accurately reflect user experience. The study also investigates the opportunities and challenges presented by cloud computing, particularly in multi-cloud environments, advocating for standardized approaches to resource management and orchestration. Auto-scaling techniques are central to this work, crucial for achieving elasticity in cloud applications by dynamically adjusting resources based on demand. The research further considers the role of fog and edge computing in extending cloud capabilities to the network edge, enabling low-latency processing, and highlights the benefits of policy-based management systems for automating cloud resource provisioning and adaptation. The integration of expert systems and AI techniques for automating SLA management and optimizing resource allocation is also explored, alongside mechanisms for resolving conflicts between policies in complex cloud environments. Standardization across cloud providers is identified as a major obstacle to multi-cloud adoption, and the research advocates for common standards to facilitate interoperability and portability. Auto-scaling is presented as a critical technique for achieving elasticity, with various algorithms and strategies discussed, including reactive and proactive approaches. Policy-based management systems are seen as essential for automating cloud resource management, while AI and expert systems offer potential for optimizing resource allocation and resolving performance issues. Managing applications across multiple cloud providers introduces significant complexity, necessitating tools and techniques to simplify the process. The research also highlights the importance of mechanisms to resolve conflicts between policies and the benefits of fog and edge computing for applications requiring low latency and localized processing. Further analysis could focus on specific SLA specification languages, comparative analysis of auto-scaling algorithms, and the security implications of auto-scaling. Exploring cost optimization in auto-scaling, conducting real-world case studies, and investigating integration with DevOps practices would also be valuable. The impact of serverless computing on SLA management and auto-scaling represents another promising area for investigation. Overall, this research provides a valuable overview of the challenges and opportunities in SLA management, auto-scaling, and cloud computing, highlighting the importance of standardization, automation, and intelligent resource management in delivering high-quality cloud services. Agreement Translation to Vendor-Neutral Metrics This research addresses a critical challenge in multi-cloud management by developing an intelligent knowledge-based system that automatically translates service-level agreements into measurable metrics across different cloud providers.

The team engineered a system capable of interpreting high-level agreements defined by users and converting them into vendor-neutral metrics, effectively bridging the gap between consumer expectations and provider capabilities. This translation process incorporates an intelligent tutoring component, providing feedback to cloud consumers and ensuring clarity in service expectations. Central to this work is the development of a specific set of vendor-neutral metrics, designed to be consistently measurable regardless of the underlying cloud infrastructure. Scientists meticulously defined these metrics and established methods for their measurement across leading cloud providers, enabling a standardized approach to performance monitoring. The system leverages message queuing technologies, specifically RabbitMQ and Apache ActiveMQ, to facilitate communication and data exchange between different cloud environments and the central knowledge base, ensuring reliable data transmission and supporting real-time monitoring of service performance. Extensive testing, using Infrastructure as a Service and Platform as a Service use cases, validated the system’s ability to automatically and transparently exploit multi-cloud resources across various application domains.

The team harnessed techniques from expert systems and Boolean expression matching, utilizing a BE-Tree index structure to efficiently manage and evaluate complex service level requirements. Endorsement from cloud experts consulted during the study further validates the practicality and effectiveness of this innovative multi-cloud management solution. SLAs Automatically Translated Across Cloud Platforms This work presents a solution to a critical challenge in cloud computing: enabling consumers to effectively utilize multi-cloud environments without being locked into a single provider. Researchers developed an intelligent knowledge-based system capable of automatically translating high-level Service Level Agreements (SLAs), defined by cloud consumers, into vendor-neutral metrics measurable across multiple cloud platforms. This system addresses the problem that each cloud provider exports a different set of low-level metrics, hindering consistent monitoring and compliance. The core of this achievement lies in a knowledge base populated with production rules that embody expert knowledge, allowing the system to accurately map SLOs to specific, measurable objectives. For example, the team demonstrated that for content-based publish/subscribe middleware, a throughput of 10,000 notifications per second correlates with CPU usage not exceeding 80%. This translation process is not static; the system incorporates a feedback loop, allowing it to learn and refine its knowledge based on observed deviations in SLOs, or through direct input from experts. Furthermore, the system can aggregate knowledge, benefiting all cloud consumers from accumulated expertise. Complementing the intelligent system, researchers defined a comprehensive repository of vendor-neutral metrics spanning Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and contextual data, abstracting the specificities of each cloud provider and offering a consistent set of measurable parameters. Analysis of publish/subscribe middleware applications revealed that throughput saturation in content-based systems is linked to CPU exhaustion, while topic-based systems are limited by memory availability, informing the mapping of high-level SLOs to low-level KPIs, such as CPU usage and memory consumption. The system’s ability to handle uncertainty and missing data further enhances its robustness and applicability in diverse cloud environments. Vendor-Neutral Metrics for Multi-Cloud Compliance This research presents a solution to the challenge of utilising multi-cloud environments effectively, addressing the current reliance on cloud consumers to manage compliance with service level agreements.

The team developed an intelligent system capable of automatically translating high-level service agreements into vendor-neutral metrics, effectively bridging the gap between consumer requirements and provider-specific measurements. This system incorporates an intelligent tutoring component, providing feedback to consumers throughout the process and simplifying complex multi-cloud management. The core achievement lies in the identification and implementation of a set of vendor-neutral metrics, applicable across leading cloud providers, and a method for measuring these metrics consistently. Validation, using both Infrastructure as a Service and Platform as a Service use cases, demonstrates the system’s ability to enable transparent and automated multi-cloud exploitation across diverse application domains, a conclusion supported by expert consultation. The researchers acknowledge that a subset of complex service mappings currently require expert definition, and suggest that collaboration with cloud providers, similar to existing migration guides, could further enhance the system’s capabilities. Future work could focus on incorporating these expert-defined mappings to broaden the scope of automated multi-cloud management. 👉 More information 🗞 From SLA to vendor-neutral metrics: An intelligent knowledge-based approach for multi-cloud SLA-based broker 🧠 ArXiv: https://arxiv.org/abs/2510.21173 Tags: