Predictive Orchestration for Elastic Cloud Resource Optimization

Cloud computing is an on-demand computing model, using virtualization technology to provide services to users in the form of virtual machines (VMs) through internet. Job-scheduling is one of the fundamental reasons to have high performance in a cloud environment. Therefore, there is a need for protocols that can efficiently allocate jobs to resources. However, due to the lack of an explicit and formal description of the resource perspective in the existing cloud platforms, the correctness of Cloud resources management can not be verified. The aim of the present work is to offer a formal description as a step towards ensuring a correct and consistent Cloud resource allocation modeling. In particular, we propose a distributed formal model for the description of a job-scheduling protocol in the cloud which considers the types of jobs and the resource availability in its scheduling decision. The formal verification of different properties of the proposed model has been performed automatically using Model-checking.

This paper addresses the impact of Virtual Memory Streaming (VMS) technique in provisioning virtual machines (VMs) in cloud environment. VMS is a scaling virtualization technology that allows different virtual machines rapid scale, high performance, and increase hardware utilization. Traditional hypervisors do not support true no-downtime live migration, and its lack of memory oversubscription can hurt the economics of a private cloud deployment by limiting the number of VMs on each host. VMS brings together several advanced hypervisor memory management techniques including granular page sharing, dynamic memory footprint management, live migration, read caching, and a unique virtual machine cloning capability. An architecture model is described, together with a proof-of-concept implementation, that VMS dynamically scaling of virtualized infrastructure with true live migration and cloning of VMs. This paper argues that VMS for Cloud allows requiring significantly reduced server memory and reducing the time for virtualized resource scaling by instantly adding more virtual machines.

The integration of machine learning (ML) models with cloud computing has transformed the landscape of predictive analytics, offering scalable, efficient, and flexible solutions for organizations. Cloud platforms such as AWS, Google Cloud, and Microsoft Azure enable businesses to deploy and manage complex ML models without the need for extensive on-premise infrastructure. However, optimizing these ML models for performance and cost-efficiency in cloud environments presents unique challenges, including resource management, latency, scalability, and data security. This paper focuses on strategies to optimize machine learning models specifically for predictive analytics in cloud environments. It explores key techniques such as auto-scaling, model compression, and hyperparameter tuning, which are critical for improving the accuracy and speed of predictions while minimizing computational costs. The research also examines advanced tools such as containerization, serverless computing, and cloud-native services that further streamline the deployment and management of ML models. In the Indian context, where cloud adoption is growing rapidly, optimizing ML models is crucial for businesses across various sectors, including finance, healthcare, and e-commerce. By leveraging cloud-based ML solutions, Indian companies can enhance their predictive analytics capabilities, driving smarter decision-making and operational efficiency. This abstract presents an overview of how optimized machine learning models can unlock the full potential of predictive analytics in cloud environments, leading to better business outcomes. Through case studies and practical applications, this paper provides actionable insights into the best practices for optimizing ML models in a cloud-based setting.

In recent decades, task scheduling and load balancing in the cloud is a growing research area, due to the vast amount of data stored in the server highly increases the load. In order to address this concern, Hybrid Max-Min Genetic Algorithm (HMMGA) is proposed for task scheduling and load balancing in the cloud environment. At first, the load is evaluated for every Virtual Machine (VM), if the load is high, then HMMGA is used for balancing the load. HMMGA selects the best VMs to assign the tasks and migrates the over-loaded VMs tasks to the under-loaded VMs. HMMGA significantly avoids the imbalanced workload performance in the cloud environment. In this research paper, the proposed HMMGA performance is compared to Max-Min algorithm, Low time complexity and low cost binary Particle Swarm Optimizer (IBPSO-LBS) and PSO with Technique of Order Preference by Similarity to Ideal Solution (TOPSIS) algorithm to examine the efficacy of HMMGA. From the experimental simulation, the result shows that HMMGA averagely delivers 1.63 and 3.88 seconds less make span compared to the Max-Min and TOPSIS-PSO algorithm for five VMs. In addition, HMMGA averagely enhances 10% to 40% of resource utilization than the MaxMin and TOPSIS-PSO algorithm. In another experiment, the HMMGA approximately showed 1.7 to 25.99 seconds less average waiting time compared to the Max-Min and IBPSO-LBS.

A family of heuristics for agent-based elastic Cloud bag-of-tasks concurrent scheduling

An improved genetic algorithm for task scheduling in the cloud environments using the priority queues: Formal verification, simulation, and statistical testing

Cloud computing is a technology that meets the needs of a vast number of users. Predicting ‎workload and scheduling are often the elements that determine cloud performance. This study ‎addresses the challenge of efficient load balancing in scalable cloud environments, where ‎traditional methods fail under dynamic workloads, leading to resource wastage and ‎performance degradation. To overcome this, an advanced framework integrating Deep ‎Learning (DL) and Reinforcement Learning (RL) is proposed. Simulation data on CPU usage, ‎memory, network traffic, and execution times are collected and preprocessed using mean ‎imputation and Min-Max normalization. A Sliding Window Approach with Multi-Scale ‎Convolutional Bidirectional LSTM (MS-Conv-BiLSTM) is used for time-series feature ‎extraction. An Attention-based LSTM forecasts workload levels, while a lightweight CNN ‎assists in task classification. Adaptive load balancing decisions are optimized using the Double Deep Q-Network (DDQN), aiming to reduce latency and response time while maximizing throughput and resource utilization. Experimental results confirm that the DL-RL framework ‎significantly enhances real-time load balancing performance in cloud environments‎.

Cloud Service Providers supply services to clients in terms of their demands. They need to be constantly under monitoring for their services with respect to consensus agreements between clients and service providers. A Third Party Auditor or TPA as a trusted organization appears to be necessary to monitor executing agreements of cloud services. Using a third party as an extra component creates cost overheads for clients in a cloud environment. Thus, introducing a cost efficient framework for a cloud environment which includes a third party is an eminent achievement to make a TPA feasible and practical in cloud environments. In this paper, we propose a TPA framework for monitoring service level agreements between cloud service providers and cloud clients using several cloud resources. This framework employs different types of service deployments from various cloud service providers excluding the cloud service provider which is being monitored. Then, we demonstrate that the framework can mitigate costs of a third party auditor in a cloud environment. Simulations of trends for costs exhibits cost efficiency of at least forty percent over ten years when a TPA follows our proposed framework in comparison to other frameworks. Finally, we provide an analysis to compare characteristics of our framework with other frameworks and discuss the advantages of our proposed framework. Our results indicate that TPA as a component of the framework not only reduces overall costs of its presentation in a cloud environment but additionally improves management efficiency and security.

In the evolving landscape of modern IT infrastructure, organizations are increasingly adopting hybrid environments that combine on-premises data centers with cloud solutions to meet scalability, flexibility, and operational efficiency needs. However, managing the security and performance of such complex environments introduces unique challenges, particularly in areas like network segmentation, vulnerability management, and real-time threat detection. This case study explores the implementation of an AI-powered network infrastructure audit and security framework for a global retail company operating in both on-premises and cloud environments. The company’s data center infrastructure relies on Cisco devices, including Cisco ACI (Application Centric Infrastructure) for network segmentation, Cisco Firepower NGFW for intrusion prevention, and Cisco Umbrella for DNS-level protection. The core of this approach involves leveraging AI-driven tools to automate vulnerability management, threat detection, and response across both on-premises systems and cloud environments (AWS, Azure). The study illustrates how Cisco ACI and AWS Security Hub were used to segment networks and isolate sensitive data, minimizing the attack surface. Furthermore, the integration of Qualys VMDR for automated vulnerability scanning and patch management, combined with Ansible for efficient patch deployment, improved operational efficiency and reduced manual intervention in vulnerability remediation. Cisco Tetration provided enhanced visibility into workload behaviors across both environments, ensuring that unauthorized lateral movements were prevented. The implementation of AI-driven security analytics through Cisco Firepower and Umbrella allowed for rapid identification and mitigation of potential threats. Additionally, the study highlights the importance of automated compliance reporting, with tools like ServiceNow integrated for tracking remediation efforts and generating compliance reports for industry standards such as PCI-DSS and GDPR. By deploying a combination of advanced network segmentation, real-time threat detection, and automated remediation, the company achieved significant improvements in its security posture, operational efficiency, and compliance. This case study serves as a comprehensive example of how AI-powered security tools and Cisco devices can be effectively integrated to address the complexities of managing a hybrid IT infrastructure while ensuring that both on-premises and cloud environments remain secure, efficient, and compliant. The results of this implementation demonstrate the value of combining cutting-edge security technologies with automation to drive proactive risk management and operational excellence

Large-scale quantum chemistry computations, such as those executed with the Tensor Algebra for Many-body Methods (TAMM) framework, require careful configuration of runtime parameters to achieve high performance and cost efficiency in high-performance computing (HPC) and cloud environments. Without effective performance analysis tools, researchers risk inefficient use of computational resources, leading to longer runtimes and higher costs. To address this challenge, this thesis presents the design and implementation of a performance profiling and visualization toolkit for TAMM, developed as part of the DOE TEC4 project in collaboration with Pacific Northwest National Laboratory, Microsoft, and Louisiana State University. The toolkit collects detailed runtime data from TAMM-based quantum chemistry computations, enabling the analysis of tensor contraction performance, visualization of dependency graphs, and estimation of compute costs through machine learning models. A user-facing dashboard provides interactive performance views and supports optimization of configuration parameters such as tile size and node allocation. The toolkit is evaluated using Coupled Cluster Singles and Double (CCSD) workloads on both LSU HPC systems and Microsoft Azure HPC cloud nodes. Results show that the machine learning models achieve accurate runtime prediction (Mean Absolute Percentage Error as low as 2.3% in certain test cases) and that the visual analytics effectively reveal performance bottlenecks. These capabilities make exascale-level simulations more accessible,interpretable, and resource-efficient for chemists and computational scientists.

Water Resource Agency (WRA) has integrated cloud computing technologies based on cloud service framework for setting up easy-to-use data exchanging portal. Due to climate change and its effects on global society, the sustainability of water resource became a significant and highly attentional issue. In Taiwan, the demands of water resource information have increased either government agency, private institution, non-profit organization or domestic. Therefore, WRA followed the trend of open data, expanded nodes of current cloud environment and upgraded the version of a management tool for keeping the high performance, large capacity and utility of the cloud environment. Cloud computing is designed to provide the platform for data storage and user-friendly application procedures and interfaces in WRA. Data security of personnel is seen as well in this paper. In addition, for proceeding with the correction and stability of Taiwan water exchanging data standard, several procedures and systems are development as counterparts that help the affairs of data examination.

Remote geovisualization has gained momentum to support large‐scale geospatial data analysis and complex decision‐making over the last few years. Cloud computing, due to its capabilities to deliver on‐demand computing resources, has been embraced to develop and deploy interactive and scalable remote geovisualization applications. However, current cloud computing frameworks do not offer a versatile resource management scheme that is readily applicable for online remote visualization services, which usually require maintaining a satisfactory service level over time under dynamic workloads. To address this gap, we propose an automatic cloud resource management approach based on a bi‐level scheduling and horizontal scaling scheme to exploit cloud resources efficiently. At the lower level, a dynamic task‐scheduling scheme using collaborative filtering techniques is proposed to allocate virtual cloud resources to execute sub‐tasks. The scheduling scheme considers spatio‐temporal patterns presented in visualization views. At the upper level, reinforcement learning is adopted to perform resource auto‐scaling based on a reward function that integrates three different facets, namely: time cost, resource cost, and service stability. The original reinforcement learning algorithm is improved in two main aspects: (1) considering the delay of resource provisioning that is common in cloud environments; and (2) using online Gaussian estimation to estimate Q values. Task scheduling and auto‐scaling interact with each other and are integrated to deliver a comprehensive and responsive resource management solution. Experimental results demonstrate that our approach outperforms several existing cloud resource management methods. The proposed approach is also applicable for other interactive visualization applications, which have similar workload characteristics and performance requirements as interactive remote geovisualization.

Cloud computing has transformed IT infrastructure by offering scalable, pay-as-you-go resources. However, efficiently allocating cloud resources under dynamic workloads while minimizing operational costs and maintaining service-level agreements (SLAs) remains a critical challenge. This paper presents a hybrid AI-driven framework for cost-efficient resource allocation in cloud environments. The proposed model integrates Long Short-Term Memory (LSTM) networks for workload forecasting, reinforcement learning (RL) for dynamic decision-making, heuristic scheduling (inspired by ACO and GA) for optimal task assignments, and economic pricing using the ERA model. A real-world data set from Google Cluster Trace is used to validate the framework. Simulation results show a forecasted CPU utilization of 10.79% with an estimated cost of ₹1.08 and one SLA violation, demonstrating the potential of hybrid AI models for real-time and cost-aware resource provisioning.

Cloud computing has become indispensable for meeting the rising demand for compute-intensive applications by providing cost-effective computational and storage resources. As reliance on cloud services increases, optimizing resource allocation is more critical than ever. This study introduces CloudProphet, an innovative machine learning-based framework designed to predict virtual machine (VM) performance in cloud environments. The approach begins with Dynamic Time Warping (DTW) to classify different types of applications based on their behavioral patterns. It further employs Pearson correlation to identify runtime metrics that are highly correlated with performance, ensuring that only the most relevant features are used in prediction models. These selected metrics are incorporated into three variations of deep learning models for evaluation: (1) an LSTM model without the use of DTW or feature selection, (2) an LSTM model with both DTW classification and selected features, and (3) a GRU model leveraging both DTW and highly correlated runtime indicators. Among these, the GRU-based approach demonstrates superior performance, achieving a remarkable 99.3% accuracy in predicting VM performance. The methodology is validated using a publicly available cloud workload dataset from GitHub and is further enhanced through real-time experimentation on live datasets to ensure practical applicability. The results highlight the robustness and accuracy of the proposed GRU model in forecasting both application types and VM behavior, making it a powerful tool for improving cloud resource management. By effectively anticipating workload demands and system performance, CloudProphet aids in reducing latency, minimizing resource wastage, and enhancing overall cloud service efficiency. This study underscores the value of combining time-series alignment techniques like DTW with correlation-based metric selection and advanced deep learning models such as GRU, ultimately offering a scalable and accurate solution for proactive and intelligent cloud performance prediction.

Multi-objective optimization oriented policy for performance and energy efficient resource allocation in Cloud environment