The orchestration of cloud computing infrastructures is challenging, considering the number, heterogeneity and dynamicity of the involved resources, along with the highly distributed nature of the applications that use them for computation and storage. Evidently, the volume of relevant monitoring data can be significant, and the ability to collect, analyze, and act on this data in real time is critical for the infrastructure’s efficient use. In this study, we introduce a novel methodology that adeptly manages the diverse, dynamic, and voluminous nature of cloud resources and the applications that they support. We use knowledge graphs to represent computing and storage resources and illustrate the relationships between them and the applications that utilize them. We then train GraphSAGE to acquire vector-based representations of the infrastructures’ properties, while preserving the structural properties of the graph. These are efficiently provided as input to two unsupervised machine learning algorithms, namely CBLOF and Isolation Forest, for the detection of storage and computing overusage events, where CBLOF demonstrates better performance across all our evaluation metrics. Following the detection of such events, we have also developed appropriate re-optimization mechanisms that ensure the performance of the served applications. Evaluated in a simulated environment, our methods demonstrate a significant advancement in anomaly detection and infrastructure optimization. The results underscore the potential of this closed-loop operation in dynamically adapting to the evolving demands of cloud infrastructures. By integrating data representation and machine learning methods with proactive management strategies, this research contributes substantially to the field of cloud computing, offering a scalable, intelligent solution for modern cloud infrastructures.

We consider the planning problem of a flexible optical network. Given the traffic matrix and the transponders’ feasible configurations that account for the physical layer, we formulate the planning problem considering both the use or not of regenerators. Demands are served for their requested rates by choosing the route, selecting the transmission configuration, breaking the transmissions in more than one connection and placing regenerators, if needed, and allocating the spectrum to them. The objective is to serve the traffic and find a solution that is Pareto optimal with respect to the maximum spectrum used and the cost (number and type) of transponders used. The problem definition and the proposed algorithms are general and applicable to flex-grid as well as fixed-grid networks. We start by presenting algorithms based on integer linear programming formulations for transparent and translucent networks (without or with regenerators) and we continue by presenting heuristic algorithms. Using input driven by transmission studies on optical orthogonal frequency-division multiplexing (OFDM)-based networks we evaluate the performance gains that can be obtained by an OFDM over a mixed line rate fixed-grid WDM optical network.

The advancements in virtualization technologies and distributed computing infrastructures have sparked the development of cloud-native applications. This is grounded in the breakdown of a monolithic application into smaller, loosely connected components, often referred to as microservices, enabling enhancements in the application’s performance, flexibility, and resilience, along with better resource utilization. When optimizing the performance of cloud-native applications, specific demands arise in terms of application latency and communication delays between microservices that are not taken into consideration by generic orchestration algorithms. In this work, we propose mechanisms for automating the allocation of computing resources to optimize the service delivery of cloud-native applications over the edge-cloud continuum. We initially introduce the problem’s Mixed Integer Linear Programming (MILP) formulation. Given the potentially overwhelming execution time for real-sized problems, we propose a greedy algorithm, which allocates resources sequentially in a best-fit manner. To further improve the performance, we introduce a multi-agent rollout mechanism that evaluates the immediate effect of decisions but also leverages the underlying greedy heuristic to simulate the decisions anticipated from other agents, encapsulating this in a Reinforcement Learning framework. This approach allows us to effectively manage the performance–execution time trade-off and enhance performance by controlling the exploration of the Rollout mechanism. This flexibility ensures that the system remains adaptive to varied scenarios, making the most of the available computational resources while still ensuring high-quality decisions.

The common practice in optical networks is to deploy them statically in a future proof approach and to operate them with only traffic/fault driven changes. Lightpaths are overprovisioned with high quality of transmission (QoT) margins to anticipate future degradations, due to interference, aging, and maintenance operations, and to account for inaccuracies in QoT estimation. Such assumptions decrease network efficiency and increase the cost. Moreover, transponders and regenerators are assigned once to demands, yielding low efficiency if traffic varies significantly. Elastic optical technology, feedback from software monitors, and software defined networking enable the dynamic operation of the network. We propose an algorithm that takes into account the actual physical layer and traffic conditions and provisions new or adapts existing lightpaths with actual (just-enough) margins, optimizing placement and transmission parameter decisions for the transponders and regenerators. Using the proposed algorithm in a multi-period scenario, we quantify the benefits that can be achieved when planning with actual margins and varying levels of traffic dynamicity.

The uncertain and non-dispatchable nature of renewable energy sources renders Demand Response (DR) a critical component of modern electricity distribution systems. Demand Response (DR) service provision takes place via aggregators and special distribution-level markets (e.g., flexibility markets), where small, distributed DR resources, such as building energy management systems, electric vehicle charging stations, micro-generation and storage, connected to the low-voltage distribution grid, offer DR services. In such systems, energy balancing (and thus, also DR decisions) have to be made close to real-time. Thus, market clearing algorithms for DR service provision must fulfill several requirements related to the efficiency of their operation. More specifically, a DR market clearing algorithm needs to be optimal in terms of cost-efficiency, scalable in terms of number of assets and locations, and able to satisfy real-time constraints. In order to cope with these challenges, this article presents a distributed DR market clearing algorithm based on Lagrangian decomposition, combined with an optimal cloud resource allocation algorithm for assigning the required computation power. A heuristic algorithm is also presented, able to achieve a near-optimal solution, within negligible computational time. Simulations, performed on a testbed, demonstrate the computational burden introduced by various DR models, as well as the heuristic algorithm's near-optimal performance. The resource allocation algorithm is able to service multiple DR requests (e.g., in multiple distribution networks), and minimize the cost of computational resources while respecting the execution time constraints of each request. This enables third parties to offer cost-efficient and competitive DR operation as a service.