EMC+: An Opportunistic Elasticity Method for Improving System Throughput and CPU Utilization in Cloud Data Centers

Enhancing the Kubernetes Scheduler: A State-of-the-Art Review from Cloud to Edge

A comprehensive review of Kubernetes scheduling strategies published between January 2023 and January 2026 is presented and a multi-dimensional taxonomy is established that categorizes scheduling approaches based on common objectives, modification methods, optimization methodologies, targeted workloads, evaluation methods, scheduling scopes, and performance metrics.

Evaluating Paramount Iterations as a Performance Proxy for High Performance Computing Applications in the Cloud

Cloud computing has transformed data storage, access, and processing, offering significant opportunities and advantages. However, selecting the most cost‐effective and efficient cloud resources for high‐performance computing (HPC) workloads remains a challenge, due to their long‐running and resource‐intensive characteristics that require tailored allocation strategies. In this work, a methodology is presented to optimize cloud resource selection for HPC applications using a performance proxy based on a few Paramount Iterations (PIs). The proxy enables the estimation of Relative Performance and Relative Cost between two cloud configurations with minimal overhead, allowing different optimization strategies to identify configurations that minimize execution time or cost. This approach extends previous work by evaluating a broader range of applications and cloud configurations, systematically identifying limitations and highlighting scenarios where the technique is most effective. The methodology was evaluated across three cloud providers, and results show that selecting instances based on the second PI achieves at least 97% of the best achievable execution time in most cases, while highlighting cases where additional PIs improve selection accuracy. These findings confirm that the proposed methodology effectively predicts optimal cloud resources for HPC workloads, reducing verification costs and providing a practical framework for informed resource selection.

A Tepid-Aware container scheduler for serverless computing environments targeting throughput and energy efficiency

Release of a Comprehensive Dataset for Fostering Dynamic Power Management Mechanisms in the Edge-Cloud Continuum

The rapid development and scaling of mobile telecommunications networks, together with related domains such as the edge-cloud continuum have raised significant concerns regarding energy consumption and environmental sustainability. Addressing these concerns requires a focus on CPU energy consumption, as CPUs are among the largest energy consumers in these systems. This paper investigates existing techniques, with a focus on CPU idle states (C-states), performance states (P-states), and frequency scaling governors implemented at both hardware and software levels. These mechanisms enable the dynamic adjustment of CPU parameters, providing opportunities to optimize power consumption, frequency, voltage, and overall system performance. In this regard, three CPUs with different architectures from well-known manufacturers, Intel® and AMD®, are thoroughly examined. A comprehensive dataset, collected under three load scenarios (idle, medium, and high), is used to support the analysis, reflect realistic runtime conditions, and enable a comparison of the technological differences in how these parameters are exposed and utilized.

CELLServe: An SLO-Aware and Cost Efficient LLMs Serving System for Serverless Computing Environments

CELLServe formalizes SLO-constrained joint resource provisioning as an optimization problem with a dedicated algorithm, and introduces an opportunistic instance merging strategy for decode phase functions to reclaim fragmented resources.

Performance modeling in FaaS workloads at the edge using machine learning

The Function as a Service (FaaS) paradigm has emerged as a compelling architectural model for both cloud and edge computing environments, enabling the execution of self-contained functions triggered by specific events while abstracting from developers infrastructure management complexities such as load balancing and auto-scaling. In FaaS-enabled clusters, particularly within resource-constrained edge environments, precise resource consumption estimation becomes critical to optimize resource utilization, minimize latency, prevent system overloads, and ensure scalability. This paper addresses performance modeling challenges in FaaS-enabled distributed and decentralized edge computing systems, operating at the granularity level of both nodes and individual functions. We propose a Machine Learning-based framework designed to predict key performance indicators, including CPU utilization, memory, and energy consumption, based on incoming workload patterns, while simultaneously forecasting potential system overload conditions. Moreover, our approach introduces a profiling methodology that characterizes serverless functions according to their resource consumption profiles, thereby enabling accurate prediction of node-level resource demands without requiring detailed knowledge of individual deployed functions. Experimental validation demonstrates that our predictive models achieve 97% accuracy in anticipating node overload scenarios, providing a robust foundation for proactive resource management in edge-based FaaS deployments. Moreover, while our best individual, function-based regression models predict node-level CPU, RAM and power consumption with a Mean Absolute Percentage Error below 9% on average, our experiments highlight the effectiveness of function profiling and cluster-based modeling. On one hand a novel multi-target regressor based on a permutation-invariant neural architecture is proved to generalize effectively across previously unseen workload compositions in the tested scenarios, with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R^2$$\end{document} scores ranging from 0.94 to 0.98, i.e., aligned with the performance of specialized single-target regression models. On the other hand, cluster-level models generalize effectively to previously unseen functions within the same usage class, maintaining prediction errors within practically acceptable ranges and often in the single-digit percentage range.