Heterogeneous Cloud Resources Research Articles

Secure Federated Learning Across Heterogeneous Cloud and High-Performance Computing Resources: A Case Study on Federated Fine-Tuning of LLaMA 2

Secure Federated Learning Across Heterogeneous Cloud and High-Performance Computing Resources: A Case Study on Federated Fine-Tuning of LLaMA 2

A multi-hierarchy particle swarm optimization-based algorithm for cloud workflow scheduling

In this paper, we consider the market-driven workflow scheduling problem on heterogeneous cloud resources with deadline constraints. The transmission delay between dependent tasks on different virtual machines (VMs) is considered. The objective is to minimize the total monetary cost, which is computed based on the on-demand price structure. Inspired by the heuristic and meta-heuristic algorithms, a multi-hierarchy particle swarm optimization (MHPSO) algorithm is proposed, which mainly consists of three components: (1) the workflow aggregation method to pretreat the workflow structure considering the transmission cost and the VM utilization; (2) the initial population generating method to create a set of initial particles with the given encoding method and the solution generation method; and (3) the hierarchical evolving process where particles are divided into multiple groups and evolved iteratively. We calibrate the parameters and components of the proposal statistically over five well-known workflows with different deadline settings. A comparison of the proposed algorithm to existing methods for the considered problem is carried out. Experimental results demonstrate the proposal is effective for the problem under study.

Adaptation of Parallel SaaS to Heterogeneous Co-Located Cloud Resources

Cloud computing has received increasing attention due to its promise of delivering on-demand, scalable, and virtually unlimited resources. However, heterogeneity or co-location of virtual cloud resources can cause severe degradation of the efficiency of parallel computations because of a priori unknown application-specific performance metrics, load imbalance, and limitations of memory bandwidth. This paper presents the runtime adaptation of parallel discrete element method (DEM) Software as a Service (SaaS) to heterogeneous or co-located resources of the OpenStack cloud. The computational workload is adapted by using weighted repartitioning and runtime measured performance of parallel computations on Docker containers. The high improvement in performance up to 48.7% of the execution time is achieved, applying the runtime adapted repartitioning when the load imbalance is high enough. The low load imbalance leads to the close values of computational load, when small variations in the system load and performance can cause oscillations in subsets of particles. Memory stress tests cause heterogeneity of non-isolated containers, which reduces the performance of memory bandwidth bound DEM SaaS on the co-located resources. The runtime adapted repartitioning handles the constant and periodically variable performance of non-isolated containers and decreases the total execution time of DEM SaaS.

FMMEHO Based Workflow Scheduling in Virtualized Cloud Environment for Smart Irrigation System

Extensive and exhaustive water utilization for agriculture, industries and ground water consumption for domestic purposes has heavily deterioted the water bodies. Cloud and sensor technology is widely deployed in a several real-time applications, especially in agriculture. The transformation of data obtained from large sensor networks into a valuable knowledge and assests for applications can effectively leverage the techniques like Cloud Computing (CC). In CC, scheduling the workflow is the major concern that focuses on comprehensive execution of workflows without compromising the Quality of Service (QoS). But workflow scheduling augmented with resource allocation is extremely challenging task because of its inherent computational intensity, task dependencies, and heterogeneous cloud resources. In this article, a novel Optimum Energy and Resource Aware Workflow Scheduling (OERES) scheme that is motivated by popular Fuzzy Membership Mutation Elephant Herding Optimization (FMMEHO) algorithm is proposed, that aims to schedule the task workflow to Virtual Machines (VMs) that are involved in computation. This also concentrates on dynamically deploying and un-deploying the VMs pertaining to the task requirements. The FMMEHO algorithm is a popular nature inspired technique, which is rooted on herding patterns of the giant mammals, the elephants. This algorithm employs a clan operator that updates the location and distance of elephants depending resource and energy usage of each clan in the context of matriarch elephant. The proposed OERES schema elevates the resource utilization and simultaneously mitigates the energy usage without compromising the dependency and deadline constraints. This work uses the famous Cloud Sim simulator to simulate the underlying cloud environment to investigate the effectiveness of proposed model. The efficacy of the scheduling methods is examined based on important parameters like mean Resource Utilization (RU), Energy utilization or Consumption/ Task (ECT), Total Energy Consumption (TEC), Makespan and Execution Time per Task (ETT). The results very well portray the effectiveness of proposed OERES algorithm against already existing methods.

A Survey on Machine Learning for Geo-Distributed Cloud Data Center Management

Cloud workloads today are typically managed in a distributed environment and processed across geographically distributed data centers. Cloud service providers have been distributing data centers globally to reduce operating costs while also improving quality of service by using intelligent workload and resource management strategies. Such large scale and complex orchestration of software workload and hardware resources remains a difficult problem to solve efficiently. Researchers and practitioners have been trying to address this problem by proposing a variety of cloud management techniques. Mathematical optimization techniques have historically been used to address cloud management issues. But these techniques are difficult to scale to geo-distributed problem sizes and have limited applicability in dynamic heterogeneous system environments, forcing cloud service providers to explore intelligent data-driven and Machine Learning (ML) based alternatives. The characterization, prediction, control, and optimization of complex, heterogeneous, and ever-changing distributed cloud resources and workloads employing ML methodologies have received much attention in recent years. In this article, we review the state-of-the-art ML techniques for the cloud data center management problem. We examine the challenges and the issues in current research focused on ML for cloud management and explore strategies for addressing these issues. We also discuss advantages and disadvantages of ML techniques presented in the recent literature and make recommendations for future research directions.

A Complexity Aware Scheduler With Dynamic Slot Allocation for Cloud Video Transcoding

For a multimedia cloud computing platform, it needs to perform video transcoding to provide bandwidth-compatible bit-streams for users. In configuring a MapReduce system, it allocates slots to workers with the assumption of homogeneous worker power and performs task scheduling by assuming equal task time complexity. However, the computing power of a practical cluster of workers and the time complexity of tasks is time-varying in any case. The task-scheduling algorithm has to well manipulate these heterogeneous cloud resources and task complexity. We first find a good partition size for a video transcoding job for efficient processing. We proposed a Complexity-Aware Scheduling (CAS) algorithm that reorders task assignment priorities according to task complexity to maintain load-balancing operations. By utilizing a Neural Network model to refine the task complexity estimation for the CAS(CASNN), the scheduling and transcoding speedup performances can further be improved. Based on the CASNN, we proposed a Dynamically Adjusting Slot number allocation (DAS) method, DASCASNN, to adjust the slots according to resource utilization status to improve the processing performance. Experimental results show the proposed DASCASNN can help reduce 30% of the transcoding time on average as compared to available schedulers and increase the resource utilization rates from 82.7% to 98%.

On the Optimization of Self-Organization and Self-Management Hardware Resource Allocation for Heterogeneous Clouds

There is a tendency, during the last years, to migrate from the traditional homogeneous clouds and centralized provisioning of resources to heterogeneous clouds with specialized hardware governed in a distributed and autonomous manner. The CloudLightning architecture proposed recently introduced a dynamic way to provision heterogeneous cloud resources, by shifting the selection of underlying resources from the end-user to the system in an efficient way. In this work, an optimized Suitability Index and assessment function are proposed, along with their theoretical analysis, for improving the computational efficiency, energy consumption, service delivery and scalability of the distributed orchestration. The effectiveness of the proposed scheme is being evaluated with the use of simulation, by comparing the optimized methods with the original approach and the traditional centralized resource management, on real and synthetic High Performance Computing applications. Finally, numerical results are presented and discussed regarding the improvements over the defined evaluation criteria.

FastTrack: Minimizing Stalls for CDN-Based Over-the-Top Video Streaming Systems

Traffic for internet video streaming has been rapidly increasing and is further expected to increase with the higher definition videos and IoT applications, such as 360 degree videos and augmented virtual reality applications. While efficient management of heterogeneous cloud resources to optimize the quality of experience is important, existing work in this problem space often left out important factors. In this paper, we present a model for describing a today’s representative system architecture for video streaming applications, typically composed of a centralized origin server and several CDN sites. Our model comprehensively considers the following factors: limited caching spaces at the CDN sites, allocation of CDN for a video request, choice of different ports from the CDN, and the central storage and bandwidth allocation. With the model, we focus on minimizing a performance metric, stall duration tail probability (SDTP), and present a novel, yet efficient, algorithm to solve the formulated optimization problem. The theoretical bounds with respect to the SDTP metric are also analyzed and presented. Our extensive simulation results demonstrate that the proposed algorithms can significantly improve the SDTP metric, compared to the baseline strategies. Small-scale video streaming system implementation in a real cloud environment further validates our results.

Scoring and Dynamic Hierarchy-Based NSGA-II for Multiobjective Workflow Scheduling in the Cloud

Cloud computing becomes a promising technology to reduce computation cost by providing users with elastic resources and application-deploying environments as a pay-per-use model. More scientific workflow applications have been moved or are being migrated to the cloud. Scheduling workflows turns to the main bottleneck for increasing resource utilization and quality of service (QoS) for users. This work formulates workflow scheduling as multiobjective optimization problems and proposes a Scoring and Dynamic Hierarchy-based NSGA-II (Nondominated Sorting Genetic Algorithm II), called SDHN for short, to minimize both makespan and cost of workflow execution. First, a scoring criterion is developed to calculate the total score for each individual during population updating, which is used as a quantitative index to evaluate the dominance degree of individuals among the whole population. Hence, SDHN can distinguish individuals within the same dominance level and target its search toward the directions of elite solutions as their different dominance degrees and accordingly improve search efficiency. Second, a population-based dynamic hierarchical structure (HS) and its evolutionary rules are presented to update HS by comparing each child with all parental individuals from bottom to up until finding a proper dominant level. Since traversing all HS levels is not needed in most cases, the number of individual comparisons is reduced and SDHN&#x2019;s updating efficiency is greatly improved, especially for large-scale and complex applications. Third, to guarantee its converging to the near-optimal solutions, adaptive adjustment strategies (AASs) are designed to prevent the search from falling into local optima or diverging by checking the number of individuals at the highest HS level and then modifying the relevant genetic operations to guide the evolutionary process to approach the global Pareto Front. Extensive experiments are conducted to verify SDHN, and the results show that it outperforms the existing algorithms in the quality and diversity of resulting solutions as well as convergence time. <i>Note to Practitioners</i>&#x2014;Most scientific applications are computation and/or data-intensive and need large-scale or high-performance resources for their execution. More and more scientists use workflows to manage their applications, but how to efficiently run them in the cloud is a big challenge due to their large scale as well as the dynamic characteristics of the elastic and heterogeneous cloud resources. In this article, we develop a novel multiobjective optimization technique for workflow scheduling such that the makespan and cost can be minimized simultaneously. A scoring criterion, dynamic hierarchical structure and its evolutionary rules, and adaptive adjustment strategies are designed to cooperate with each other and increase the search ability and efficiency of the original and widely used NSGA-II. Adequate experiments are conducted to verify the proposed method&#x2019;s performance, and the experimental results show that it can provide more near-optimal solutions than the existing methods. It can be readily applied for implementing more efficient and effective cloud data centers to execute large-scale scientific workflows.

Hybrid Workflow Scheduling on Edge Cloud Computing Systems

Internet of Things applications can be represented as workflows in which stream and batch processing are combined to accomplish data analytics objectives in many application domains such as smart home, health care, bioinformatics, astronomy, and education. The main challenge of this combination is the differentiation of service quality constraints between batch and stream computations. Stream processing is highly latency-sensitive while batch processing is more likely resource-intensive. In this work, we propose an end-to-end hybrid workflow scheduling on an edge cloud system as a two-stage framework. In the first stage, we propose a resource estimation algorithm based on a linear optimization approach, gradient descent search (GDS), and in the second stage, we propose a cluster-based provisioning and scheduling technique for hybrid workflows on heterogeneous edge cloud resources. We provide a multi-objective optimization model for execution time and monetary cost under constraints of deadline and throughput. Results demonstrate the framework performance in controlling the execution of hybrid workflows by efficiently tuning several parameters including stream arrival rate, processing throughput, and workflow complexity. In comparison to a meta-heuristics technique using Particle Swarm Optimization (PSO), the proposed scheduler provides significant improvement for large-scale hybrid workflows in terms of execution time and cost with an average of 8% and 35%, respectively.

Unequal‐interval based loosely coupled control method for auto‐scaling heterogeneous cloud resources for web applications

SummaryMost existing quality of service (QoS) control algorithms of Web applications take into account Web Server or database connections which can be released immediately. However, many applications are deployed on virtual machines (VMs) or even Spot VMs elastically rented from public Clouds. To save costs, interval‐priced VMs are not released until the ends of rented intervals. Such delays of control effects make existing methods rent or release excess VMs leading to overcontrol. Fluctuated prices make Spot VMs unreliable due to unexpected termination which makes fault‐tolerant strategies crucial. In this article, an unequal‐interval‐based loosely coupled control method is proposed to improve the quality of service (QoS) control ability of fault‐tolerant strategies. A queuing model with arrival‐rate‐adjustment coefficient is used to predict required capacity as a feedforward controller. Another two‐threshold and queuing‐model‐based method is applied to update the coefficient as a loosely coupled feedback controller. Meanwhile, unequal‐interval controller collaborating method is proposed to avoid overcontrol and react quickly to workload changes. Our approach is evaluated on both a simulation platform and a real Kubernetes Cluster. Experimental results illustrate that our approach decreases the percentage of waiting times larger than service level agreements with similar or lower rental costs compared with existing algorithms.

GRP-HEFT: A Budget-Constrained Resource Provisioning Scheme for Workflow Scheduling in IaaS Clouds

In Infrastructure as a Service (IaaS) Clouds, users are charged to utilize cloud services according to a pay-per-use model. If users intend to run their workflow applications on cloud resources within a specific budget, they have to adjust their demands for cloud resources with respect to this budget. Although several scheduling approaches have introduced solutions to optimize the makespan of workflows on a set of heterogeneous IaaS cloud resources within a certain budget, the hourly-based cost model of some well-known cloud providers (e.g., Amazon EC2 Cloud) can easily lead to a higher makespan and some schedulers may not find any feasible solution. In this article, we propose a novel resource provisioning mechanism and a workflow scheduling algorithm, named Greedy Resource Provisioning and modified HEFT (GRP-HEFT), for minimizing the makespan of a given workflow subject to a budget constraint for the hourly-based cost model of modern IaaS clouds. As a resource provisioning mechanism, we propose a greedy algorithm which lists the instance types according to their efficiency rate. For our scheduler, we modified the HEFT algorithm to consider a budget limit. GRP-HEFT is compared against state-of-the-art workflow scheduling techniques, including MOACS (Multi-Objective Ant Colony System), PSO (Particle Swarm Optimization), and GA (Genetic Algorithm). The experimental results demonstrate that GRP-HEFT outperforms GA, PSO, and MOACS for several well-known scientific workflow applications for different problem sizes on average by 13.64, 19.77, and 11.69 percent, respectively. Also in terms of time complexity, GRP-HEFT outperforms GA, PSO and MOACS.

Online scheduling of dependent tasks of cloud’s workflows to enhance resource utilization and reduce the makespan using multiple reinforcement learning-based agents

Due to different heterogeneous cloud resources and diverse and complex applications of the users, an optimal task scheduling, which can satisfy users and cloud service providers with energy-saving and cost-effective use of resources, is a major issue in cloud computing. On the one hand, network users are demanding the quality assurance of their requested services, minimizing their costs, and their own data security, and on the other hand, the service providers consider less power consumption, more efficient use of resources, and optimal utilization. In dependent tasks dealing with massive data, resource scheduling is considered as an important challenge. Due to the time limitation of online scheduling process of dependent tasks, many existing methods of the literature are not able to guarantee the best resource utilization. In this paper, a reinforcement learning approach is exploited in a multi-agent system for task scheduling and resource provisioning, in order to reduce the makespan, minimize the required power, optimize the cost of using the resources, and maximize the utilization of the resources (considering their expiration time), simultaneously. The proposed algorithm has two phases. In the first phase, the tasks are scheduled using reinforcement learning techniques, and in the second one, considering the information obtained from the scheduling phase, resources are allocated in a multi-agent environment. The results of experiments show that this method improves the efficiency of the use of resources and reduces their costs. Moreover, the expiration time of the tasks is observed and the total execution time and energy consumption will be significantly reduced.

Frequency-Minimal Utility-Maximal Moving Target Defense Against DDoS in SDN-Based Systems

With the increase of DDoS attacks, resource adaptation schemes need to be effective to protect critical cloud-hosted applications. Specifically, they need to be adaptable to attack behavior, and be dynamic in terms of resource utilization. In this paper, we propose an intelligent strategy for proactive and reactive application migration by leveraging the concept of ‘moving target defense’ (MTD). The novelty of our approach lies in: (a) stochastic proactive migration frequency minimization across heterogeneous cloud resources to optimize migration management overheads, (b) market-driven migration location selection during proactive migration to optimize resource utilization, cloud service providers (CSPs) cost and user quality of experience, and (c) fast converging cost-minimizing reactive migration coupled with a ‘false reality’ pretense to reduce the future attack success probability. We evaluate the effectiveness of our proposed MTD-based defense strategy using a Software-defined Networking (SDN) enabled GENI Cloud testbed for a “Just-in-time news articles and video feeds” application. Our frequency minimization results show more than 40% reduction in DDoS attack success rate in the best cases when compared to the traditional periodic migration schemes on homogeneous cloud resources. The results also show that our market-driven migration location selection strategy decreases CSP cost and increases resource utilization by 30%.

Load Balancing Algorithm for Efficient VM Allocation in Heterogeneous Cloud

Cloud computing is a growing service computing trend that offers users a range of on-demand variety of services from applications, processing capability, and storage based on the concept of the “Pay-As-PerUse” model. Organizations from every sector are now realizing the benefits offered by cloud computing technology and moving towards the cloud. Cloud computing offers numerous advantages over conventional computing. However, it still faces a few challenges such as resource utilization in a cloud data centre and quality of service to the end-users due to improper workload balance among available resources. Heterogeneous cloud resources also impact cloud systems overall performance. We proposed an enhanced load balancing algorithm in this research paper for efficient VM allocation in a heterogeneous cloud. Our proposed algorithm allocates independent user tasks or requests to available virtual machines in cloud datacentre efficiently to manage proper load balancing. This algorithm is aimed at minimizing user request response time and the time required for data centre processing. The results obtained showed a significant reduction in user request response time and data centre processing time as compared to “Throttled” and “Round Robin (RR)” algorithms.

Proactive VNF Scaling with Heterogeneous Cloud Resources: Fusing Long Short-Term Memory Prediction and Cooperative Allocation

Network function virtualization (NFV) is designed to implement network functions by software that replaces proprietary hardware devices in traditional networks. In response to the growing demand of resource-intensive services, for NFV cloud service providers, software-oriented network functions face a number of challenges, such as dynamic deployment of virtual network functions and efficient allocation of multiple resources. This study aims at the dynamic allocation and adjustment of network multiresources and multitype flows for NFV. First, to seek a proactive approach to provision new instances for overloaded VNFs ahead of time, a model called long short-term memory recurrent neural network (LSTM RNN) is proposed to estimate flows in this paper. Then, based on the estimated flow, a cooperative and complementary resource allocation algorithm is designed to reduce resource fragmentation and improve the utilization. The final results demonstrate the advantage of the LSTM model on predicting the network function flow requirements, and our algorithm achieves good results and performance improvement in dynamically expanding network functions and improving resource utilization.

Load Balancing Algorithm for Efficient VM Allocation in Heterogeneous Cloud

Cloud computing is a growing service computing trend that offers users a range of on-demand variety of services from applications, processing capability, and storage based on the concept of the “Pay-As-PerUse” model. Organizations from every sector are now realizing the benefits offered by cloud computing technology and moving towards the cloud. Cloud computing offers numerous advantages over conventional computing. However, it still faces a few challenges such as resource utilization in a cloud data centre and quality of service to the end-users due to improper workload balance among available resources. Heterogeneous cloud resources also impact cloud systems overall performance. We proposed an enhanced load balancing algorithm in this research paper for efficient VM allocation in a heterogeneous cloud. Our proposed algorithm allocates independent user tasks or requests to available virtual machines in cloud datacentre efficiently to manage proper load balancing. This algorithm is aimed at minimizing user request response time and the time required for data centre processing. The results obtained showed a significant reduction in user request response time and data centre processing time as compared to “Throttled” and “Round Robin (RR)” algorithms.

An efficient resource deployment method for steam- based stochastic demands in distributed cloud platforms

It has been a consensus that deploying geographically dispersed stream-based online services into distributed cloud platforms has gained exceptional advantages. Globally visiting services make user requests characterised with dramatic fluctuation, which introduces stochastic demands for various resources. In order to maximise satisfied user requests and guarantee quality-of-service under given expense budget, efficient resource deployment becomes the key to this problem. We propose a stochastic demand oriented resource deployment method with more profits and less time complexity. Experiments using simulated and realistic data indicate that proposed method can outperform existing algorithms by increasing the weighted summation of satisfied demands up to 37%, fit for all scenarios with heterogeneous distributed cloud resources.

An efficient resource deployment method for stream-based stochastic demands in distributed cloud platforms

It has been a consensus that deploying geographically dispersed stream-based online services into distributed cloud platforms has gained exceptional advantages. Globally visiting services make user requests characterised with dramatic fluctuation, which introduces stochastic demands for various resources. In order to maximise satisfied user requests and guarantee quality-of-service under given expense budget, efficient resource deployment becomes the key to this problem. We propose a stochastic demand oriented resource deployment method with more profits and less time complexity. Experiments using simulated and realistic data indicate that proposed method can outperform existing algorithms by increasing the weighted summation of satisfied demands up to 37%, fit for all scenarios with heterogeneous distributed cloud resources.

Adaptive Provisioning of Heterogeneous Cloud Resources for Big Data Processing

Efficient utilization of resources plays an important role in the performance of large scale task processing. In cases where heterogeneous types of resources are used within the same application, it is hard to achieve good utilization of all of the different types of resources. By taking advantage of recent developments in cloud infrastructure that enable the use of dynamic clusters of resources, and by dynamically altering the size of the available resources for all the different resource types, the overall utilization of resources, however, can be improved. Starting from this premise, this paper discusses a solution that aims to provide a generic algorithm to estimate the desired ratios of instance processing tasks as well as ratios of the resources that are used by these instances, without the necessity for trial runs or a priori knowledge of the execution steps. These ratios are then used as part of an adaptive system that is able to reconfigure itself to maximize utilization. To verify the solution, a reference framework which adaptively manages clusters of functionally different VMs to host a calculation scenario is implemented. Experiments are conducted based on a compute-heavy use case in which the probability of underground pipeline failures is determined based on the settlement of soils. These experiments show that the solution is capable of eliminating large amounts of under-utilization, resulting in increased throughput and lower lead times.