An Empirical Study of Task Scheduling Strategies for Image Processing Application on Heterogeneous Distributed Computing System

Loosely coupled applications composed of a potentially very large number (from tens of thousands to even billions) of tasks are commonly used in high-throughput computing and many-task computing paradigms. To efficiently execute large-scale computations which can exceed the capability in a single type of computing resources within expected time, we should be able to effectively integrate resources from heterogeneous distributed computing (HDC) systems such as clusters, grids, and clouds. In this paper, we quantitatively analyze the performance of three different real scientific applications consisting of many tasks on top of HDC systems based on a partnership of distributed computing clusters, grids, and clouds to understand the application and resource characteristics, and show practical issues that normal scientific users can face during the course of leveraging these systems. Our experimental study shows that the performance of a loosely coupled application can be significantly affected by the characteristics of a HDC system, along with hardware specification of a node, and their impacts on the performance can vary widely depending on the resource usage pattern of each application. We then devise a preference-based scheduling algorithm that can reflect characteristics and resource usage patterns of various loosely coupled applications running on top of HDC systems from our experimental study. Our preference-based scheduling algorithm can allocate the resources from different HDC systems to loosely coupled applications based on the preferences of the applications for the HDC systems. We evaluate the overall system performance over various preference types, using trace-based simulations, which can be determined based on different factors such as CPU specifications and application throughputs. Our simulation results demonstrate the importance of understanding the application and resource characteristics on effective scheduling of loosely coupled applications on the HDC systems.

The paper presents a task allocation technique for multiple applications onto heterogeneous distributed computing system to minimize the overall makespan. An existing critical path based algorithm for scheduling of tasks of single application has been used to allocate tasks of multiple applications onto heterogeneous distributed computing system. The paper discusses how a composite application is given for multiple applications and how critical path based algorithm is applied efficiently for the allocation of tasks of different applications.

Load balancing and task partitioning are important components of distributed computing. The optimum performance from the distributed computing system is achieved by using effective scheduling and load balancing strategy. Researchers have well explored CPU, memory, and I/O-intensive tasks scheduling, and load balancing techniques. But one of the main obstacles of the load balancing technique leads to the ignorance of applications having a mixed nature of tasks. This is because load balancing strategies developed for one kind of job nature are not effective for the other kind of job nature. We have proposed a load balancing scheme in this paper, which is known as Mixed Task Load Balancing (MTLB) for Cluster of Workstation (CW) systems. In our proposed MTLB strategy, pre-tasks are assigned to each worker by the master to eliminate the worker's idle time. A main feature of MTLB strategy is to eradicate the inevitable selection of workers. Furthermore, the proposed MTLB strategy employs Three Resources Consideration (TRC) for load balancing (CPU, Memory, and I/O). The proposed MTLB strategy has removed the overheads of previously proposed strategies. The measured results show that MTLB strategy has a significant improvement in performance.

Previously, DAG scheduling schemes used the mean (average) of computation or communication time in dealing with temporal heterogeneity. However, it is not optimal to consider only the means of computation and communication times in DAG scheduling on a temporally (and spatially) heterogeneous distributed computing system. In this paper, it is proposed that the second order moments of computation and communication times, such as the standard deviations, be taken into account in addition to their means, in scheduling “stochastic” DAGs. An effective scheduling approach which accurately estimates the earliest start time of each node and derives a schedule leading to a shorter average parallel execution time has been developed. Through an extensive computer simulation, it has been shown that a significant improvement (reduction) in the average parallel execution times of stochastic DAGs can be achieved by the proposed approach.

Executing complicated computations in parallel increases the speed of computing and brings user delight to the system. Decomposing the program into several small programs and running multiple parallel processors are modeled by Directed Acyclic Graph. Scheduling nodes to execute this task graph is an important problem that will speed up computations. Since task scheduling in this graph belongs to NP-hard problems, various algorithms were developed for node scheduling to contribute to quality service delivery. The present study brought a heuristic algorithm named looking ahead sequencing algorithm (LASA) to cope with static scheduling in heterogeneous distributed computing systems with the intention of minimizing the schedule length of the user application. In the algorithm proposed here, looking ahead is considered as a criterion for prioritizing tasks. Also, a property called Emphasized Processor has been added to the algorithm to emphasize the task execution on a particular processor. The effectiveness of the algorithm was shown on few workflow type applications and the results of the algorithm implementation were compared with two more heuristic and meta-heuristic algorithms.

It is the need of an era to store and process big data and its applications. To process these applications, it is inevitable to use heterogeneous distributed computing systems (HeDCS). The heterogeneous distributed systems facilitate scalability, an essential characteristic for big data processing. However, to implement the scalable model, it is essential to handle performance, efficiency, optimal resource utilization and several other key constraints. Scheduling algorithms play a vital role in achieving better performance and high throughput in heterogeneous distributed computing systems. Hence, selection of a proper scheduling algorithm, for the specific application, becomes a critical task. Selection of an appropriate scheduling algorithm in heterogeneous distributed computing systems require the consideration of various parameters like scheduling type, multi-core processors, and heterogeneity. The paper discusses broadly the hierarchical classification of scheduling algorithms implemented in heterogeneous distributed computing systems and presents a comparative study of these algorithms, thus providing an insight into the significance of various parameters that play a role in the selection of a scheduling algorithm.

With the development of high‐performance computing, it is possible to solve large‐scale computing problems. However, the irregularity and access characteristics of computing problems bring challenges to the realisation and performance optimisation. Improving the performance of a single core makes it challenging to maintain Moore's law, and multi‐core processors emerge. A chip brings together multiple universal processor cores of equal status and has the same structure supported by an isomorphic multi‐core processor. In high‐performance computing, the granularity of computing tasks leads to the complexity of scheduling strategies. Satisfying high system performance, load balancing and processor fault tolerance at a minimum cost is the key to task scheduling in the high‐performance field, especially in specific multi‐core hardware architecture. In this study, global real‐time task scheduling is implemented in a high‐performance multi‐core system. The system adopts the hybrid scheduling among clusters and the intelligent fitting within clusters to implement the global real‐time task scheduling strategy. In the cluster scheduling policy, tasks are allowed to preempt the core with low priority, and the priority of tasks that access memory is dynamically improved, higher than that of all the tasks without memory access. An intelligent fitting method is also proposed. When the data read by the task is in the cache and the cache access ability value of the task is within a reasonable threshold, the priority of the task is promoted to the highest priority, preempting the core without the access memory task. The results show that the intelligently fitting global scheduling strategy for multi‐core systems has better performance in the nuclear utilisation rate and task schedulability.

A hybrid heuristic–genetic algorithm for task scheduling in heterogeneous processor networks

In the Spark distributed framework, data communication problems (network transfer overhead, network IO bottlenecks) caused by data transfer across nodes/racks are a common cause of performance degradation due to the inconsistency between the task execution location and the data location. Additionally, in heterogeneous environments, Spark's task scheduling strategy cannot fully utilize the advantages of high‐performance nodes. To address the above issues, firstly, this paper proposes a cost‐aware task selection strategy. The strategy models the cost of tasks by considering the impact of data locality and heterogeneous factors on the efficiency of job execution. For scenarios where data locality needs to be reduced for scheduling tasks, the task scheduling problem is transformed into a minimum weighted bipartite graph matching problem, and a greedy matching algorithm is used to solve for the minimum processing cost option. For scenarios that maintain the current data localization level for scheduling tasks, select the task execution with the largest change in task processing cost due to data localization changes. Secondly, the problem is that Spark's delay scheduling algorithm causes resources in the cluster to be in an unnecessary waiting state and reduces cluster resource utilization. In this paper, we propose an adaptive adjustment strategy for delay waiting time based on benefit assessment. This policy improves the resource utilization of the cluster by evaluating the benefit of delay waiting of the scheduler and dynamically adjusting the delay time based on the result of the evaluation. Finally, we implement the proposed strategy in Spark 3.0.0 and evaluate its performance using some representative benchmarks. The experimental results show that, compared with other task scheduling algorithms, the strategy proposed in this paper can effectively improve the execution efficiency of jobs, reduce the execution time of jobs by 15.8%–31.9%, and at the same time reduce the network traffic and improve the CPU utilization.

A New Method for Scheduling Divisible Data on a Heterogeneous Two-Levels Hierarchical System

As a main distributed computing system, Spark has been used to solve problems with more and more complex tasks. However, the native scheduling strategy of Spark assumes it works on a homogenized cluster, which is not so effective when it comes to heterogeneous cluster. The aim of this study is looking for a more effective strategy to schedule tasks and adding it to the source code of Spark. After investigating Spark scheduling principles and mechanisms, we developed a stratifying algorithm and a node scheduling algorithm are proposed in this paper to optimize the native scheduling strategy of Spark. In this new strategy, the static level of nodes is calculated, the dynamic factors such as the length of running tasks, and CPU usage of work nodes are considered comprehensively. And through a series of comparative experiments in alienation cluster, the new strategy costs less running time and lower CPU usage rate than the original Spark strategy, which verifies that the new schedule strategy is more effective one.

In a multithreaded execution environment, race condition leads to computational errors and system hazards. Up to date, a series of task scheduling strategies have been presented in the literature to reduce the risk of race condition. Because of the increasing complexity of this problem in multi-core systems, existing task scheduling approaches are not very efficient. To deal with this challenge, in this work, we develop a race-condition-aware and hardware-oriented task partitioning and scheduling algorithm using entropy maximization model. We model uncertainty as a probabilistic occurrence within a time interval, and hence the characteristics of event ordering are analyzed through an uncertainty matrix. Next, a metric is developed to measure the uncertainty of task execution in various execution environments. Finally, a maximum entropy model is generated to ensure the lowest probability of race condition during task execution. The smallest one among maximum entropy values is chosen and used in our proposed task scheduling algorithm. Experimental results show that the proposed task scheduling strategy based on our maximum entropy model outperforms existing state-of-the-art approaches. For example, in a 128-core computing system, the task execution time, CPU utilization ratio, and throughput of our proposed task scheduling is improved by <inline-formula> <tex-math notation="LaTeX">$15.3\sim 36.4$</tex-math></inline-formula> percent, <inline-formula><tex-math notation="LaTeX">$8.2\sim 17.6$</tex-math> </inline-formula> percent, and <inline-formula><tex-math notation="LaTeX">$20.7\sim 41.4$</tex-math></inline-formula> percent, respectively. Moreover, our proposed scheduling algorithm exhibits low computational complexity and good adaptivity to diverse execution environments.

In order to improve the task scheduling strategy, a method based on genetic algorithm in cloud computing environment was proposed. First, the independent task scheduling algorithm and associated task scheduling algorithm commonly used in cloud computing are studied and compared, respectively, and their application characteristics, advantages, and disadvantages are analyzed in detail. Second, an independent task scheduling strategy based on multipopulation genetic algorithm is proposed for independent task scheduling in cloud environment, considering the scheduling time, scheduling cost, and system resource utilization of task set. The implementation steps of the algorithm are given in detail. Finally, the simulation experiment is carried out on Cloud Sim platform. The experimental results show that computing resource M is 10, subtask N is 2000, population size S is 80, and ETC matrix and RCU array are randomly generated by the system. As the number of iterations increases, the scheduling scheme formed by MCGA and CGA is more obvious and close to the subtask execution cost optimization. Finally, the optimized scheme is basically formed. However, the scheduling scheme formed by TGA has no obvious optimization effect on the subtask execution cost. It is proved that the algorithm proposed in this paper can effectively optimize the task scheduling efficiency and improve the utilization of cloud computing resources at the same time, providing a feasible idea and method for task scheduling in the cloud computing environment.

Heterogeneous systems with multiple different compute devices have come into common use recently, and the heterogeneity of the compute device is mainly reflected in three aspects: hardware architecture, instruction set architecture, and processing capability. Heterogeneous CPU-accelerator systems have attracted increasing attention especially. To make full use of multiple CPUs and accelerators to execute data-parallel applications, programmers may need to manually map computation and data to all available compute devices, which is tedious, error-prone, and difficult. Especially for some data-parallel applications, the inter-device communication could easily become the performance bottleneck of multi-device co-execution. Therefore, firstly, a runtime system is designed for supporting heterogeneous cooperative execution (HCE) of data-parallel applications, which can help programmers to automatically and efficiently map computation and data to multiple compute devices. Secondly, an incremental data transfer method is designed to avoid redundant data transfers between devices, and a three-way overlapping communication optimization method based on software pipelining is designed to effectively hide the inter-device communication overhead. Based on our previously proposed feedback-based dynamic and elastic task scheduling (FDETS) scheme and asynchronous-based dynamic and elastic task scheduling (ADETS) scheme, the modified FDETS that supports incremental data transfer and the modified ADETS that supports three-way overlapping communication optimization are proposed, which not only can effectively partition and balance the workload among multiple compute devices but also can significantly reduce data transfer overhead between devices. Thirdly, a prototype of the proposed runtime system is implemented, which provides a set of runtime APIs for task scheduling, device management, memory management, and transfer optimization. Our experimental results show that the communication overhead between devices is greatly reduced using the proposed inter-device communication optimization methods and the multi-device co-execution significantly outperforms the best single-device execution.

This paper addresses the problem of static load balancing in heterogeneous distributed computing systems taking into account both memory and communication capacity constraints. It first models the load balancing problem as an optimization problem. It then presents a modified genetic algorithm, called Adaptive Genetic Algorithm (AGA), to solve the problem. The performance of the proposed algorithm is evaluated by simulation studies on randomly generated instances and the results are compared with that obtained by applying both the Genetic Algorithm (GA) and the Simulated Annealing (SA). Also, the qualities of the results are compared with the optimal solutions that obtained by applying the Brach-and-Bound (BB) algorithm.