Job Completion Time Research Articles

Geo-Distributed IoT Data Analytics With Deadline Constraints Across Network Edge

Owing to the advancement of the Internet of Things (IoT) and 5G mobile technologies, various IoT devices produce massive data, which is usually transferred to nearby sites, such as edge nodes or datacenters. Many large-scale IoT applications need to analyze the data distributed across multiple sites to obtain final results. A dominant challenge of this type of data analytics is the heterogeneities of resource capacities across geo-distributed sites. In this article, we find that the resource capacity as well as the resource price differ among sites, and the price heterogeneity has a significant impact on geo-distributed IoT data analytics. Thus, each geo-distributed IoT data analytics job prefers to minimize the job execution cost while guaranteeing its deadline requirement under the resource constraints of involved sites. Specifically, we propose to jointly consider the resource heterogeneities of both capacity and price, and minimize the cost of each job before its deadline. We characterize this optimization problem as a quadratically constrained quadratic programming problem. To tackle such an NP-hard problem, we propose the minimize the job completion cost before a given deadline (MCGL) method, which calculates a task placement solution by the gradient adjustment strategy according to the remarkable negative correlation relationship between job completion time and job completion cost of geo-distributed IoT data analytics job. The task placement strategy can optimize resource cost with respect to the deadline requirement of any geo-distributed data analytics job. The trace-driven evaluations indicate that MCGL significantly reduces the total cost compared with existing methods; moreover, they satisfy the deadline constraints simultaneously.

ASTRAEA: A Fair Deep Learning Scheduler for Multi-Tenant GPU Clusters

Modern GPU clusters are designed to support distributed Deep Learning jobs from multiple tenants concurrently. Each tenant may have varied and dynamic resource demands. Unfortunately, existing GPU schedulers fail to thoroughly consider the fairness among the tenants and jobs, which can result in unbalanced resource allocation and unfair user experience. In this article, we present an efficient solution to provide strong fairness while maintaining high scheduling effectiveness in multi-tenant GPU clusters. First, we introduce a novel Long-Term GPU-time Fairness metric, which can comprehensively evaluate the fairness at both the tenant and job levels, based on both the temporal and spatial impacts of resource allocation. Second, we design a new and practical GPU scheduler, <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Astraea</small> , to enforce the desired fairness among tenants and jobs. Large-scale evaluations show that <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Astraea</small> can improve tenant fairness by up to 9.42× compared to state-of-the-art schedulers, without sacrificing the average job completion time.

Solving a stochastic hierarchical scheduling problem by VNS-based metaheuristic with locally assisted algorithms

High-end equipment manufacture has features of intensive technology, uncertain process, and large scale. Hence, the scheduling of the High-end equipment manufacture is challenging. Large ships are a type of High-end equipment that is tied to national maritime security. In this paper, we study a stochastic hierarchical scheduling problem with serial-batch processing machines and the deterioration effect in ship manufacture. In the studied problem, the job processing time is uncertain and follows a normal probability distribution. Under the deterioration effect, the actual processing time of jobs is formulated as a function of the start time. The first objective is to maximize the robustness of the manufacturing system. The second objective is to minimize the total inventory cost subjected to the optimization of the first objective. The robustness of the manufacturing system is measured by the worst level of completion probability. The total inventory cost is measured by the difference between the job delivery time and the job completion time. The mathematical model for the studied problem is formulated. Some structural properties are derived to solve the problem. Based on the properties, the optimal sequence for the first objective can be obtained and a heuristic algorithm is designed to optimize the second objective. The considered problem is NP-hard involving several non-linearities. Hence, a variable neighborhood search (VNS) algorithm is developed to solve the problem in a reasonable time, which is integrated with three local search strategies and a heuristic algorithm. Various problem instances are generated and extensive computational experiments verify that the proposed VNS algorithm outperforms other compared algorithms and has strong robustness.

Heuristic algorithms for carton production scheduling problems with technicians and different machines

The carton manufacturing process has several characteristics, including processing procedures, variation in type and processing procedure for each carton, specific operating machines set up by a technician, and multiple machines with different efficiencies in each workstation. Thus, decisions on pairing technicians, machines, and products and the job processing sequence production schedule arrangement have considerable influence on the efficiency of the machines and the completion time of jobs. This study establishes a mathematical model and presents an immune-based algorithm (IBA) and genetic-based algorithm (GBA) to develop a comprised assignment decision and production schedule and determine the best decisions to efficiently minimize the total tardiness cost. The proposed conversion scheme is guaranteed to produce feasible decisions considering the relationship between the encoded value of the adapted evolutionary algorithms and decisions associated with assignments and the production schedule. Numerical examples of a real case in Taiwan are solved using the proposed IBA, GBA, and CPLEX optimization solver. Computational results show that, except for cases with a few orders, the IBA method performed best, followed by the GBA method and then the CPLEX solver.

Stability of a schedule minimising the makespan for processing jobs on identical machines

ABSTRACT A set of jobs has to be processed on identical machines. Every job may be processed on any available machine without preemptions. The criterion is to minimise the makespan (i.e. the completion time of the last job in a schedule). During the realisation of a schedule, durations of some jobs may deviate from the initial values estimated before scheduling. Other jobs have fixed durations that are known before scheduling. We conduct a stability analysis of the optimal semi-active schedule. First, we derive necessary and sufficient conditions for an optimal schedule to be unstable with respect to infinitely small variations of the non-fixed durations (the stability radius of an unstable schedule is equal to zero). Second, we show that the stability radius of an optimal schedule could be infinitely large. Furthermore, several lower and upper bounds on the stability radius have been established. Third, we derive a formula and develop an algorithm for calculating stability radii.

Load Balancing Algorithms for Hadoop Cluster in Unbalanced Environment.

Considering that in the process of job scheduling, the cluster load should be prebalanced rather than remedied when the load is seriously unbalanced; therefore, in this paper, the task scheduling flow of the Hadoop cluster is analyzed deeply. On the Hadoop platform, a self-dividing algorithm is proposed for load balancing. An intelligent optimization algorithm is used to solve load balance. A dynamic feedback load balancing scheduling method is proposed from the point of view of task scheduling. In order to solve the shortcoming of the fair scheduling algorithm, this paper proposes two ways to improve the resource utilization and overall performance of Hadoop. When the mapping task is completed and the tasks to be reduced are assigned, the task assignment is based on the performance of the nodes to be reduced. It gives full play to the advantages of the ant colony algorithm and the hive colony algorithm so that the fusion algorithm can better deal with load balance. Then, three existing scheduling algorithms are introduced in detail: single queue scheduling, capacity scheduling, and fair scheduling. On this basis, an improved task scheduling strategy based on genetic algorithm is proposed to allocate and execute application tasks to reduce task completion time. The experiment verifies the effectiveness of the algorithm. The LBNP algorithm greatly improves the efficiency of reducing task execution and job execution. The delay capacity scheduling algorithm can ensure that most tasks can achieve localization scheduling, improve resource utilization, improve load balance, and speed up job completion time.

A coupling optimization method of production scheduling and computation offloading for intelligent workshops with cloud-edge-terminal architecture

With the rapid development and application of Industrial Internet of Things (IIoT) technology, the amount of data generated during the operations of intelligent workshops shows an explosive growth trend. Real-time analysis and processing of these data is the premise of rapid intelligent decision-making. Therefore, the cloud-edge-terminal (CET) architecture is usually adopted in manufacturing sites, and the overall delay of data processing can be reduced by designing a reasonable offloading strategy for computing tasks. In general, computing tasks are generated by demand-driven monitoring, measurement, analysis, and decision-making during the executions of production jobs. There is a strong coupling relationship between them, and the scheduling scheme of production jobs largely determines the generation sequence and processing requirements of computing tasks. However, the current optimization methods for production scheduling and computation offloading are carried out separately to pursue their own goals, resulting in that the balance between production efficiency and computing delay cannot be achieved. To this end, this paper proposes a Production scheduling-Computation offloading Coupling Optimization (PCCO) model for intelligent workshops with CET architecture for the first time, which takes the maximum completion time of production jobs and the minimum total offloading delay time of computing tasks as dual optimization objectives. Then a four-layer coding method is designed and an enhanced multi-population multi-objective whale optimization algorithm (EMMOWOA) is proposed to solve the PCCO model. In EMMOWOA, the arithmetic exploration mechanism is adopted to strengthen the exploration ability, and a new leader selection method is proposed to lead the search of population to avoid prematurely trapping into the local optimum; to avoid the loss of diversity, an adaptive multi-population strategy is adopted. A series of comparative experimental results based on multi-objective standard test functions show that the proposed EMMOWOA has better convergence and diversity than the newly proposed multi-objective optimization algorithms, which proves the effectiveness of the proposed algorithm improvement strategies. Finally, simulation cases and a practical production case prove that the proposed coupling optimization method can balance production efficiency and computing delay well, and better meet the optimization requirements of intelligent manufacturing sites.

Theoretical Approaches Parallel Identical Machines with Multi-Objective Functions

In this study, we propose multi-objective functions which consist of the sum of completion time, tardiness time and earliness time where Ci denoted the completion time of job (i), Ti=max{Ci-di,0}, denotes the tardiness of job (i), Ei=max{di-Ci,0} be denoted the earliness of job (i).This problem is defined by // In this paper, we will present some theoretical analysis discussion, and prove when we have a problem with scheduling n jobs on two identical parallel machines (IPMSP).

Forseti: Dynamic chunk-level reshaping for data processing on heterogeneous clusters

Data-intensive computing frameworks typically split job workload into fixed-size chunks, allowing them to be processed as parallel tasks on distributed machines. Ideally, when the machines are homogeneous and have identical speed, chunks of equal size would finish processing at the same time. However, such determinism in processing time cannot be guaranteed in practice. Diverging processing times can result from various sources such as system dynamics, machine heterogeneity, and variable network conditions. Such variation, together with dynamics and uncertainty during task processing, can lead to significant performance degradation at job level, due to long tails in job completion time resulted from residual chunk workload and stragglers.In this paper, we propose Forseti, a novel processing scheme that is able to reshape data chunk size on the fly with respect to heterogeneous machines and a dynamic execution environment. Forseti mitigates residual workload and stragglers to achieve significant improvement in performance. We note that Forseti is a fully online scheme and does not require any a priori knowledge of the machine configuration nor job statistics. Instead, it infers such information and adjusts data chunk sizes at runtime, making the solution robust even in environments with high volatility. In its implementation, Forseti also exploits a virtual machine reuse feature to avoid task start-up and initialization cost associated with launching new tasks. We prototype Forseti on a real-world cluster and evaluate its performance using several realistic benchmarks. The results show that Forseti outperforms a number of baselines, including default Hadoop by up to 68% and SkewTune by up to 50% in terms of average job completion time.

Speculative Container Scheduling for Deep Learning Applications in a Kubernetes Cluster

In the past decade, we have witnessed a dramatically increasing volume of data collected from varied sources. The explosion of data has transformed the world as more information is available for collection and analysis than ever before. To maximize the utilization, various machine and deep learning models have been developed, e.g. CNN [1] and RNN [2], to study data and extract valuable information from different perspectives. While data-driven applications improve countless products, training models for hyperparameter tuning is still a time-consuming and resource-intensive process. Cloud computing provides infrastructure support for the training of deep learning applications. The cloud service providers, such as Amazon Web Services [3], create an isolated virtual environment (virtual machines and containers) for clients, who share physical resources, e.g., CPU and memory. On the cloud, resource management schemes are implemented to enable better sharing among users and boost the system-wide performance. However, general scheduling approaches, such as spread priority and balanced resource schedulers, do not work well with deep learning workloads. In this project, we propose SpeCon, a novel container scheduler that is optimized for shortlived deep learning applications. Based on virtualized containers, such as Kubernetes [4] and Docker [5], SpeCon analyzes the common characteristics of training processes. We design a suite of algorithms to monitor the progress of the training and speculatively migrate the slow-growing models to release resources for fast-growing ones. Specifically, the extensive experiments demonstrate that SpeCon improves the completion time of an individual job by up to 41.5%, 14.8% system-wide and 24.7% in terms of makespan.

Preemptive Scheduling for Distributed Machine Learning Jobs in Edge-Cloud Networks

Recent advances in 5G and edge computing enable rapid development and deployment of edge-cloud systems, which are ideal for delay-sensitive machine learning (ML) applications such as autonomous driving and smart city. Distributed ML jobs often need to train a large model with enormous datasets, which can only be handled by deploying a distributed set of workers in an edge-cloud system. One common approach is to employ a parameter server (PS) architecture, in which training is carried out at multiple workers, while PSs are used for aggregation and model updates. In this architecture, one of the fundamental challenges is how to dispatch ML jobs to workers and PSs such that the average job completion time (JCT) can be minimized. In this work, we propose a novel online preemptive scheduling framework to decide the location and the execution time window of concurrent workers and PSs upon each job arrival. Specifically, our proposed scheduling framework consists of: i) a job dispatching and scheduling algorithm that assigns each ML job to workers and decides the schedule to train each data chunk; ii) a PS assignment algorithm that determines the placement of PS. We prove theoretically that our proposed algorithm is <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$D_{max}(1+1/\epsilon)$ </tex-math></inline-formula> -competitive with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$(1 + \epsilon)$ </tex-math></inline-formula> -speed augmentation, where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$D_{max}$ </tex-math></inline-formula> is the maximal number of data chunks in any job. Extensive testbed experiments and trace-driven simulations show that our algorithm can reduce the average JCT by up to 30% compared with state-of-the-art baselines.

Co-Scheduler: A Coflow-Aware Data-Parallel Job Scheduler in Hybrid Electrical/Optical Datacenter Networks

To support higher demand for datacenter networks, networking researchers have proposed hybrid electrical/optical datacenter networks (Hybrid-DCN) that leverages optical circuit switching (OCS) along with traditional electrical packet switching (EPS). However, due to the high <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">reconfiguration</i> delay of OCS, OCS is used only for bulk data transfers between racks to amortize the reconfiguration delay. Existing job schedulers for data-parallel frameworks are not designed for Hybrid-DCN, since they neither place tasks to aggregate data traffic to take advantage of OCS, nor schedule tasks to minimize the Coflow completion time (CCT). In this paper, we describe the mismatch between existing job schedulers and the advanced Hybrid-DCN, introduce the requirements for the new scheduler, and present the implementation of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Co-scheduler</i> , a job scheduler for data-parallel frameworks that aims to improve job performance by placing the tasks of jobs to aggregate enough data traffic to better leverage OCS to minimize the CCT in Hybrid-DCN. Specifically, for every job, Co-scheduler computes <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">guidelines</i> on how many racks to place the job’s input data and the job’s tasks. The guidelines are dynamically generated based on the real-time job characteristics or predictable job characteristics from prior runs, with the aim of leveraging OCS whenever possible and efficient and minimizing CCT of jobs. Co-scheduler then schedules the tasks of jobs based on the guidelines. We evaluate the effectiveness of Co-scheduler using trace-driven simulation. The evaluation demonstrates that Co-scheduler can improve makespan, average job completion time, and average CCT of a workload by up to 56%, 61%, and 79%, respectively, compared to the state-of-the-art schedulers.

Unrelated Parallel Machine Scheduling with Job Splitting, Setup Time, Learning Effect, Processing Cost and Machine Eligibility

This work investigates an unrelated parallel machine scheduling problem in the shared manufacturing environment. Based on practical production complexity, five job and machine-related factors, including job splitting, setup time, learning effect, processing cost and machine eligibility constraint, are integrated into the considered problem. Parallel machines with uniform speed but non-identical processing capabilities are shared on a sharing service platform, and jobs with different types can only be processed by the machines with matching eligibilities. The platform pays an amount of processing cost for using any machine to process the jobs. To balance the processing cost paid and the satisfaction of customers, we aim to minimize the weighted sum of total processing cost and total completion time of jobs in the considered problem. We establish a mixed integer linear programming model, and provide a lower bound by relaxing the machine eligibility constraint. The CPLEX solver is employed to generate optimal solutions for small-scale instances. For large-scale instances, we propose an efficient heuristic algorithm. Experimental results demonstrate that for various instance settings, the proposed algorithm can always produce near optimal solutions. We further present several managerial insights for the shared manufacturing platform.

SMART: Speedup Job Completion Time by Scheduling Reduce Tasks

SMART: Speedup Job Completion Time by Scheduling Reduce Tasks

Scheduling on uniform machines with a conflict graph: complexity and resolution

AbstractThis paper deals with the problem of scheduling a set of unit‐time jobs on a set of uniform machines. The jobs are subject to conflict constraints modeled by a graph G called the conflict graph, in which adjacent jobs cannot be processed on a same machine. The objective considered herein is the minimization of maximum job completion time in the schedule, which is famous to be NP‐hard in the strong sense. The first part of this paper is an extensive study of the computational complexity of the problem restricted to several graph classes, namely: split graphs, interval graphs, forests, trees, paths and cycles. Afterward, we focus on the resolution of the problem with arbitrary conflict graphs. For this latter, a combination of a mixed integer linear programming (MILP) formulation, lower and upper bounds is proposed. A wild range of computational experiments proved the efficiency of this technique to tremendously reduce runtime and produce more optimal solutions (around 80% in average). Furthermore, a deep analysis of the resolution process based on both the density of the conflict graph as well as machine speeds (including identical machines) is thoroughly reported.

Load balancing technique based on hybrid resource utilization in cloud computing

Cloud computing uses the internet to supply dynamic services including memory, data, bandwidth and applications. Work schedules have an influence on cloud service reliability and performance. A proper provisioning method is required for a systematic resource allocation, which comprises of large virtual resources. Depending on the present state of the system, load balancing solutions can be distinguished as dynamic or static. Dynamic or static load balancing solutions can be employed to increase server response time or to raise load balancing factors for quicker and more efficient resource utilization. To decrease the load across resources and maximize CPU usage, a hybrid load balancing technique is developed. In the cloud, we have a finite quantity of resources that must be efficiently managed in order to fulfill tasks. Requests are transmitted to a cloud server, which assigns work via quadratic probing. During load balancing, the load is shifted from heavy-weighted servers to lighter-weighted servers, enhancing CPU usage. The suggested methodology's performance was assessed using average mean response time, make-span, average make-span, and average resource utilization. The Load Balancing Algorithm (LBA) is created with the primary purpose of reducing job completion time and increasing the average resource utilization ratio.

Memetic algorithm for unrelated parallel machine scheduling problem with grey processing times

PurposeThis paper aims to provide a promising memetic algorithm (MA) for an unrelated parallel machine scheduling problem with grey processing times by using a simple dispatching rule in the local search phase of the proposed MA.Design/methodology/approachThis paper proposes a MA for an unrelated parallel machine scheduling problem where the objective is to minimize the sum of weighted completion times of jobs with uncertain processing times. In the optimal schedule of the problem’s single machine version with deterministic processing time, the machine has a sequence where jobs are ordered in their increasing order of weighted processing times. The author adapts this property to some of their local search mechanisms that are required to assure the local optimality of the solution generated by the proposed MA. To show the efficiency of the proposed algorithm, this study uses other local search methods in the MA within this experiment. The uncertainty of processing times is expressed with grey numbers.FindingsExperimental study shows that the MA with the swap-based local search and the weighted shortest processing time (WSPT) dispatching rule outperforms other MA alternatives with swap-based and insertion-based local searches without that dispatching rule.Originality/valueA promising and effective MA with the WSPT dispatching rule is designed and applied to unrelated parallel machine scheduling problems where the objective is to minimize the sum of the weighted completion times of jobs with grey processing time.

Joint optimization of production and maintenance for a serial–parallel hybrid two-stage production system

Effective machine maintenance and job scheduling strategies are two critical interactive tasks to upgrade the production efficiency in manufacturing industries. In this study, we mainly consider the interaction between production scheduling and maintenance, while optimizing the job scheduling problem in a serial–parallel hybrid two-stage production system. We formulate the problem by introducing the stochastic and workload-ratio degradation, and the actual job completion time and maintenance costs can be modeled according to the degradation status of each machine. To solve this problem, a two-step opportunistic maintenance (OM) strategy is developed, and finally an adaptive random-key genetic algorithm (GA) is designed by incorporating the OM coefficients. Numerical studies have been conducted to validate the proposed approach, and the results indicate the effectiveness compared to three existing methods under varying machine degradation scenarios.

Balanced Nonadaptive Redundancy Scheduling

Distributed computing systems implement redundancy to reduce the job completion time and variability. Despite a large body of work about computing redundancy, the analytical performance evaluation of redundancy techniques in queuing systems is still an open problem. In this work, we take one step forward to analyze the performance of scheduling policies in systems with redundancy. In particular, we study the pattern of shared servers among replicas of different jobs. To this end, we employ combinatorics and graph theory and define and derive performance indicators using the statistics of the overlaps. We consider two classical nonadaptive scheduling policies: random and round-robin. We then propose a scheduling policy based on combinatorial block designs. Compared with conventional scheduling, the proposed scheduling improves the performance indicators. We study the expansion property of the graphs associated with round-robin and block design-based policies. It turns out the superior performance of the block design-based policy results from better expansion properties of its associated graph. As indicated by the performance indicators, the simulation results show that the block design-based policy outperforms random and round-robin scheduling in different scenarios. Specifically, it reduces the average waiting time in the queue to up to 25% compared to the random policy and up to 100% compared to the round-robin policy.

An Optimization Method for Enterprise Resource Integration Based on Improved Particle Swarm Optimization.

An enterprise's development and growth are inextricably linked to rational and efficient resource integration and optimization. This study focuses on the reorganization and integration of industrial elements inside the firm from the standpoint of resource integration. The ideal resource integration strategy is investigated by integrating the industrial parts of a certain enterprise in order to increase the efficiency of project completion and lower enterprise expenses. The enterprise's internal material and human resources are limited, but it is frequently necessary to execute numerous activities at the same time, and each activity must meet multiple goals. This research investigates how to properly integrate and schedule resources while attaining different goals. This research proposes using an enhanced particle swarm optimization technique (IPSO) to combine firms' internal resources. In order to address the issue of uneven particle dispersion caused by random population initialization, IPSO incorporates chaos theory into particle population initialization. The logistic mapping sequence generates a huge number of particles, and the particles with the highest quality are chosen for initialization. This can increase particle quality, allowing particles to be spread equally during setup. In the late stage, the classic particle swarm optimization algorithm (PSO) has a slow convergence rate, causing the algorithm to readily slip into a local optimal solution. This research proposes a dynamic inertia weight update approach based on fitness value. In the later stages of the algorithm, this strategy can improve the convergence speed and quality of the global optimal solution, allowing the particles to do a global search and eventually identify the population's ideal solution. Furthermore, IPSO creates a fitness function depending on task completion time. IPSO is used to test the performance of an enterprise's resource integration case. Experiments show that the method utilized can swiftly locate the ideal solution, complete the integration, and optimization of enterprise resources in the shortest job completion time, and for the least amount of money.