Intelligent Dynamic Multi-Dimensional Heterogeneous Resource Scheduling Optimization Strategy Based on Kubernetes

PurposeIn recent years, it is imperative to establish the structure of manufacturing industry in the context of smart factory. Due to rising demand for exchange of information with various devices, and huge number of sensor nodes, the industrial wireless networks (IWNs) face network congestion and inefficient task scheduling. For this purpose, software-defined network (SDN) is the emerging technology for IWNs, which is integrated into cognitive industrial Internet of things for dynamic task scheduling in the context of industry 4.0.Design/methodology/approachIn this paper, the authors present SDN based dynamic resource management and scheduling (DRMS) for effective devising of the resource utilization, scheduling, and hence successful transmission in a congested medium. Moreover, the earliest deadline first (EDF) algorithm is introduced in authors’ proposed work for the following criteria’s to reduce the congestion in the network and to optimize the packet loss.FindingsThe result shows that the proposed work improves the success ratio versus resource usage probability and number of nodes versus successful joint ratio. At last, the proposed method outperforms the existing myopic algorithms in terms of query response time, energy consumption and success ratio (packet delivery) versus number of increasing nodes, respectively.Originality/valueThe authors proposed a priority based scheduling between the devices and it is done by the EDF approach. Therefore, the proposed work reduces the network delay time and minimizes the overall energy efficiency.

Aiming at the problem of limited resources and operational time combat air defense combat, naval combat formation analysis dynamic resources scheduling needs, on the basis of the establishment of ship dynamic scheduling model of warship formation combat resources dynamic combat formation resource scheduling, the operational resources and operational objectives, relationship between reasonable allocation of resources to play the warship formation combat operations the maximum efficiency. The design of three kinds of air defense command models: centralized platform free attack mode, command mode, request attack mode of air defense strategy, put forward the dynamic scheduling hybrid algorithm for solving the model, according to the naval fleet air defense combat scenario example for simulation. The simulation results show that the air defense strategy with the request attack mode can have effects on multi-target attacks under the shared combat situation, and the hybrid algorithm based on combat resource dynamic scheduling can meet the real-time operational requirements.

In this paper, we propose READYS, a reinforcement learning algorithm for the dynamic scheduling of computations modeled as a Directed Acyclic Graph (DAGs). Our goal is to develop a scheduling algorithm in which allocation and scheduling decisions are made at runtime, based on the state of the system, as performed in runtime systems such as StarPU or ParSEC. Reinforcement Learning is a natural candidate to achieve this task, since its general principle is to build step by step a strategy that, given the state of the system (the state of the resources and a view of the ready tasks and their successors in our case), makes a decision to optimize a global criterion. Moreover, the use of Reinforcement Learning is natural in a context where the duration of tasks (and communications) is stochastic. We propose READYS that combines Graph Convolutional Networks (GCN) with an Actor-Critic Algorithm (A2C): it builds an adaptive representation of the scheduling problem on the fly and learns a scheduling strategy, aiming at minimizing the makespan. A crucial point is that READYS builds a general scheduling strategy which is neither limited to only one specific application or task graph nor one particular problem size, and that can be used to schedule any DAG. We focus on different types of task graphs originating from linear algebra factorization kernels (CHOLESKY, LU, QR) and we consider heterogeneous platforms made of a few CPUs and GPUs. We first propose to analyze the performance of READYS when learning is performed on a given (platform, kernel, problem size) combination. Using simulations, we show that the scheduling agent obtains performances very similar or even superior to algorithms from the literature, and that it is especially powerful when the scheduling environment contains a lot of uncertainty. We additionally demonstrate that our agent exhibits very promising generalization capabilities. To the best of our knowledge, this is the first paper which shows that reinforcement learning can really be used for dynamic DAG scheduling on heterogeneous resources.

W-CDMA is the strongest candidate as the air interface technology of the third generation mobile communication systems, which are expected to support multimedia services with QoS requirements. In this paper, we review the W-CDMA technology and consider the issues in multimedia support. We propose the dynamic resource scheduling mechanism for QoS provisioning in W-CDMA through optimal power assignment and code hopping. Our framework is able to carry variable QoS multimedia traffic, efficiently serving a wide range of classes with CBR, VBR, ABR and UBR characteristics. Dynamic resource scheduling improves capacity and throughput performance. Other advantages are power saving and faster convergence to the desired power level.

<span>The distributed fog node network and variable workloads make task distribution difficult in fog computing. Optimizing computing resources for dynamic workloads with heuristic and metaheuristic algorithms has shown potential. To address changing workloads, these algorithms enable real-time decision-making. This systematic review examines heuristic, meta-heuristic, and real-time dynamic job scheduling strategies in fog computing. Static methods like heuristic and meta-heuristic algorithms can help modify dynamic task scheduling in fog computing situations. This paper covers a current study area that stresses real-time approaches, meta-heuristics, and fog computing environments' dynamic nature. It also helps build reliable and scalable fog computing systems by spotting dynamic task scheduling trends, patterns, and issues. This study summarizes and analyzes the latest fog computing research on task-scheduling algorithms and their pros and cons to adequately address their issues. Fog computing task scheduling strategies are detailed and classified using a technical taxonomy. This work promises to improve system performance, resource utilization, and fog computing settings. The work also identifies fog computing job scheduling innovations and improvements. It reveals the strengths and weaknesses of present techniques, paving the way for fog computing research to address unresolved difficulties and anticipate future challenges.</span>

The scheduling of tasks with deadlines on clusters is a key issue for offering quality-of-service (QoS) assurance. A critical challenge in real-time task scheduling is to handle various types of applications. This paper investigates the scheduling problem for processing a set of tasks comprising both divisible and indivisible real-time tasks on cluster systems. Indivisible tasks are characterized by the property that they need to be processed on their entirety on a single processor while divisible tasks can be distributed across several processing nodes by exploiting the underlying data parallelism. We propose a dynamic (on-line) real-time scheduling algorithm referred to as Hybrid Loads Push-Pull Scheduling (HLPPS) algorithm for handling a set of tasks comprising both divisible and indivisible real-time tasks on cluster systems. HLPPS is shown to efficiently exploit the parallelism in divisible tasks without undermining the schedulability of indivisible tasks and thereby optimize the overall performance. We consider two distinct network platforms - tightly coupled and loosely coupled clusters in designing the strategy. We conduct extensive performance evaluation studies to quantify the performance of the proposed algorithm under a variety of scenarios.

Task scheduling is an essential component of any distributed system because it routes tasks to appropriate resources for execution, such as grids, clouds, and peer-to-peer networks. Common scheduling algorithms include downsides, such as high temporal complexity, non-simultaneous processing of input tasks, and longer program execution times. Exploration-based scheduling algorithms prioritize tasks using a variety of methods, resulting in long execution times on heterogeneous distributed computing systems. As a result, task prioritization becomes a bottleneck in such systems. It is appropriate to prioritize tasks with the shortest execution time using faster algorithms. The genetic algorithm (GA) is one of the evolutionary approaches used to solve complex problems quickly. This paper proposes a parallel GA with a MapReduce architecture for scheduling jobs on cloud computing with various priority queues. The fundamental aim of this study is to employ a MapReduce architecture to minimize the total execution time of the task scheduling process in the cloud computing environment. The proposed method accomplishes task scheduling in two stages: first, the GA was used in conjunction with heuristic techniques to assign tasks to processors, and then the GA was used in conjunction with the MapReduce framework to assign jobs to processors. In our experiments, we consider heterogeneous resources that differ in their ability to execute various tasks, as well as running a job on different resources with varying execution durations. The results show that the proposed method outperforms other algorithms such as particle swarm optimization, whale optimization algorithm, moth-flame optimization, and intelligent water drops.

The rapid expansion of real-time, latency-sensitive Internet-of-Things (IoT) applications has revealed the limits of centralized cloud infrastructures and driven computation toward a hierarchical IoT–fog–cloud continuum. Scheduling in this environment is challenging due to resource heterogeneity, dynamic arrivals and deadlines, and competing objectives such as latency, energy, and cost. This paper introduces Quantum-inspired Biased Dynamic Scheduler (QBDS), a novel scheduling framework that optimizes a configurable Composite Objective Function (COF) combining makespan, total energy consumption, total cost, load balance, resource utilization, and temporal metrics (waiting and response times). QBDS uses (1) a priority-aware task ranking that adaptively weights deadline slack, execution length, memory footprint and data size to form a globally informed execution order; (2) a sinusoidal, quantum-inspired biasing mechanism that perturbs normalized task and node metrics via randomized mixing weights and sine modulation to escape local optima and encourage exploration of underused resources; and (3) a penalty-aware multi-objective cost evaluator for assignment decisions. Extensive experiments, including ablation studies and comparisons with state-of-the-art metaheuristics and classical heuristics, demonstrate that QBDS consistently improves makespan, energy, cost, and resource utilization across diverse workloads and topologies, while scaling robustly under heavy load.

MapReduce is now a significant parallel processing model for large-scale data-intensive applications using clusters with commodity hardware. Scheduling of jobs and tasks, and identification of TaskTrackers which are slow in Hadoop clusters are the focus research in the recent years. MapReduce performance is currently limited by its default scheduler, which does not adapt well in heterogeneous environments. In this paper, we propose a scheduling method to identify the TaskTrackers which are running slowly in map and reduce phases of the MapReduce framework in a heterogeneous Hadoop cluster. The proposed method is integrated with the MapReduce default scheduling algorithm. The performance of this method is compared with the unmodified MapReduce default scheduler. We observe that the proposed approach shows improvements in performance to the default scheduler in the heterogeneous environments. Performance improvement was observed as the overall job execution times for different workloads from HiBench benchmark suite were reduced.

Coarse-grained reconfigurable architectures (CGRAs) are increasingly employed as domain-specific accelerators due to their efficiency and flexibility. A CGRA typically relies on compilers to perform task scheduling. The longstanding problem of static scheduling is that it suffers from insufficient parallelism in handling irregularities due to over-serialization and workload imbalance, which leads to severe resource underutilization and performance loss. To counteract the limitations of static scheduling in CGRAs, it is essential to exploit dynamic parallelism automatically and manage hardware resources adaptively. However, existing dynamic scheduling mechanisms, e.g., work stealing, often reschedule aggressively for instant performance but sacrifice efficiency, which is unfavorable to CGRAs that emphasize efficiency and fewer reconfigurations. This article proposes an elastic task scheduling scheme that enables lightweight dynamic scheduling in CGRAs. Tasks are rescheduled at runtime according to the classic tagged-token dataflow paradigm to enable dynamic task-level parallelism. Meanwhile, tasks are dynamically resized according to run-time throughputs via duplication, combination, and substitution operators for balanced multitask execution. We implement the elastic task scheduling scheme on a well-known reconfigurable architecture - triggered instruction architecture (TIA). Evaluation on the MachSuite benchmarks shows that the proposed scheme is effective in improving performance and energy efficiency. The average speedup is 2× over the baseline. Also, our design attains a 57 percent improvement in the area-normalized performance and a 49 percent better energy efficiency. Compared with a state-of-the-art dynamic scheduling method, our scheme achieves 1.6× speedup and 1.6× energy efficiency than work-stealing mechanism on the same substrate.

This article describes the implementation of a dynamic scheduler for loading distribution between a hardware accelerator RTL and a CPU software task. The basic composition of a Xilinx-Zynq SoC device is a processing system (PS), coupled with FPGA programmable logic (PL). The two sections are connected via a number of Advanced eXtensible Interfaces. Hardware accelerators are mechanisms whereby different software algorithms are implemented register transfer logic (RTL) in the PL module. These accelerators determine an increased processing speed. In this article, we present a dynamic scheduler used for distribution of the load between the host processor and the RTL accelerator. There are situations in which even with increased processing speed of the accelerator, it cannot cope with the flow of data coming from memory system (shared memory). Therefore, it is necessary for this accelerator to be “aided” by a software module running in a CPU in the PS section. The article describes a scheduler that checks whether a hardware module for data processing meets the requirement of Hard Real Time (data are processed within a well-defined time frame), and in case it does not, it activates a software thread running on a CPU to support the hardware thread (out of the whole amount of data to be processed by the RTL thread, some of it is processed by the SW thread. Thus, the RTL thread will have less data to process). The scheduler activates the SW thread only when the system has to respond in real time and the amount of data cannot be processed within a certain time. Thus, the scheduler detects the need to activate the software thread that “helps” the hardware thread to process the data. The scheduler self-adjusts so that it executes a number of instructions in the software thread at all times, without introducing delays in running the RTL thread which is much faster. For this project PYNQ Z2 board, Vivado 2018.3 and Jupyter Notebook tools have been used.

The optical network integrated computing environment has been thought of as a promising technology to support large-scale data-intensive distributed computing applications. For such an environment involving so many heterogeneous resources, such as high-performance processors and optical links, faults seem to be inevitable. The faults will lead to the failure of the applications or highly delay the applications' finish times. Therefore, it is necessary to analyze resources' fault probability and then to better schedule the tasks of the application onto the appropriate resources so as to minimize the fault probability of the application. We address the task-scheduling problem based on the fault probability analysis for distributed computing applications over an optical network. We quantitatively analyze the fault probability of the processors and optical links in a given interval and propose a minimal fault probability (MFP) task-scheduling algorithm to minimize the fault probability of the application. We develop a simulator to evaluate the performance of the MFP algorithm. The simulation results prove the efficiency of the MFP algorithm.

Cloud manufacturing (CMfg) converts the traditional manufacturing system into an Internet-of-things-enabled (IoT-enabled) manufacturing system, where both manufacturing and computational tasks must be scheduled among distributed and heterogeneous resources. Deep reinforcement learning (DRL) has recently become a promising idea for task scheduling in CMfg. However, existing DRL-based methods depend heavily on problem-specific reward engineering and struggle to represent hybrid decision variables. To this end, this paper proposed the sparse-reward deep reinforcement learning (SDRL) method to solve the hybrid task scheduling problem in CMfg. First, the hybrid task scheduling model in CMfg is constructed to minimize the makespan. We reformulate the studied problem as a partially observable Markov decision process (POMDP). Then, the objective hindsight experience replay (objective HER) mechanism is proposed to alleviate the sparse reward issue, through which the scheduling policy can be effectively trained without problem-specific reward engineering. The continuous action space is defined to represent hybrid decision variables, and the implicit action-selection mapping is utilized to alleviate the boundary effect. Numerical experiments validated the effectiveness and superiority of our method compared to eleven popular scheduling algorithms including evolutionary algorithms and DRL. Compared to mainstream DRL scheduling methods, the proposed SDRL outperforms the second-best one at most by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$23.6\%$</tex-math> </inline-formula> regarding generalization, and a scheduling solution can be generated in <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.5$</tex-math> </inline-formula> seconds. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —With the intelligentization of the CMfg platform, hybrid tasks, including manufacturing and computational tasks, need to be scheduled simultaneously. However, this hybrid task scheduling problem is rarely considered by existing works. DRL exhibits many benefits in addressing scheduling problems, but the strong dependency on problem-specific reward engineering limits its application. Additionally, most DRL-based scheduling algorithms are discrete-action DRL, restricting their capacity to effectively represent hybrid decision variables. The studied problem originates from the CMfg platform, but the proposed method holds potential for broader application. The scheduling framework and the POMDP modeling can be applied to similar problems, including hybrid, manufacturing, or computational task scheduling problems. The proposed objective HER serves as a general approach to addressing challenges associated with sparse rewards, which can be extended to diverse combinatorial optimization problems aimed at optimizing an objective. We will open-source our codes to help others to apply the method to other fields.

In addition to cutting down on energy use, the green cloud-based service also drastically cuts down on operating expenses. Strongly coupled data centers demand controlled energy, constant performance, and overall optimization of excess energy consumption in order to do considerable calculations, which in turn give entire processing influence from a wide collection of resources. An energy-saving scheduling method is the focus of the research, which makes use of green cloud technologies. In recent years, cloud computing has been integrated into several domains, including computing, research, industry, and business. The need for specialized hardware along with additional resources is rendered obsolete by the provision of many services across the internet through cloud computing. Problems with energy efficiency, resource heterogeneity, and resource usage are only a few of the issues that cloud computing systems encounter. Consolidation approaches like as task scheduling as well as virtual machines (VMs) are used to address these concerns. There has been a plethora of research on task scheduling. Researchers have examined the issue using a variety of metrics and aims. The essay delves into the topic of virtualized cloud data centers' energy usage and effective resource use. Task categorization and thresholds form the basis of the proposed method, which aims to improve resource use and scheduling efficiency. Every company needs a highly efficient structure that is both flexible and homogeneous across all of their cloud environments. This research aims to reduce energy consumption in green cloud systems by using a hybrid scheduling strategy that combines a minimal completion time with a priority-based weighted round-robin. An Artificial Neural Network (ANN) based load balancing method is presented in this study. ANN makes demand predictions and then distributes resources accordingly. This method uses less energy than the cautious over-provisioning strategy since it constantly adjusts the number of active servers based on current demand. In addition, a server that is operating at a high utilization may handle more task with the same amount of power, but it will use more power overall. At last, we take a look at the current state of load balancing in cloud computing and compare it to others using different metrics.

The question of fairness has been raised in many different areas, e.g. in decision support systems, legal issues or economics. In scheduling tasks, it is not easy to know when scheduling can be called fair.We have examined previous rostering approaches to this issue, and experience shows no general formulation of fairness. Different task-specific solutions are used, or equity is defined for more straightforward, homogeneous resource scheduling. In practice, however, more complex assignments of heterogeneous groups are more common.There is also a need for a more general formulation of fairness in scheduling issues. In order to answer this, we give a basic concept for heterogeneous resources, taking into account disjoint and non-disjoint skill-sets.