Job Scheduling Policies Research Articles

A comparative study on resource allocation and energy efficient job scheduling strategies in large-scale parallel computing systems

In the large-scale parallel computing environment, resource allocation and energy efficient techniques are required to deliver the quality of services (QoS) and to reduce the operational cost of the system. Because the cost of the energy consumption in the environment is a dominant part of the owner's and user's budget. However, when considering energy efficiency, resource allocation strategies become more difficult, and QoS (i.e., queue time and response time) may violate. This paper therefore is a comparative study on job scheduling in large-scale parallel systems to: (a) minimize the queue time, response time, and energy consumption and (b) maximize the overall system utilization. We compare thirteen job scheduling policies to analyze their behavior. A set of job scheduling policies includes (a) priority-based, (b) first fit, (c) backfilling, and (d) window-based policies. All of the policies are extensively simulated and compared. For the simulation, a real data center workload comprised of 22385 jobs is used. Based on results of their performance, we incorporate energy efficiency in three policies i.e., (1) best result producer, (2) average result producer, and (3) worst result producer. We analyze the (a) queue time, (b) response time, (c) slowdown ratio, and (d) energy consumption to evaluate the policies. Moreover, we present a comprehensive workload characterization for optimizing system's performance and for scheduler design. Major workload characteristics including (a) Narrow, (b) Wide, (c) Short, and (d) Long jobs are characterized for detailed analysis of the schedulers' performance. This study highlights the strengths and weakness of various job scheduling polices and helps to choose an appropriate job scheduling policy in a given scenario.

Network and data location aware approach for simultaneous job scheduling and data replication in large-scale data grid environments

Data Grid integrates graphically distributed resources for solving data intensive scientific applications. Effective scheduling in Grid can reduce the amount of data transferred among nodes by submitting a job to a node, where most of the requested data files are available. Scheduling is a traditional problem in parallel and distributed system. However, due to special issues and goals of Grid, traditional approach is not effective in this environment any more. Therefore, it is necessary to propose methods specialized for this kind of parallel and distributed system. Another solution is to use a data replication strategy to create multiple copies of files and store them in convenient locations to shorten file access times. To utilize the above two concepts, in this paper we develop a job scheduling policy, called hierarchical job scheduling strategy (HJSS), and a dynamic data replication strategy, called advanced dynamic hierarchical replication strategy (ADHRS), to improve the data access efficiencies in a hierarchical Data Grid. HJSS uses hierarchical scheduling to reduce the search time for an appropriate computing node. It considers network characteristics, number of jobs waiting in queue, file locations, and disk read speed of storage drive at data sources. Moreover, due to the limited storage capacity, a good replica replacement algorithm is needed. We present a novel replacement strategy which deletes files in two steps when free space is not enough for the new replica: first, it deletes those files with minimum time for transferring. Second, if space is still insufficient then it considers the last time the replica was requested, number of access, size of replica and file transfer time. The simulation results show that our proposed algorithm has better performance in comparison with other algorithms in terms of job execution time, number of intercommunications, number of replications, hit ratio, computing resource usage and storage usage.

A Threshold-based Dynamic Data Replication and Parallel Job Scheduling strategy to enhance Data Grid

Data Grids provide environment for huge, data-intensive applications that produce and process enormous data. Such environments are thus asked to manage data and schedule jobs at the same time. These two important operations have to be tightly coupled to achieve the best results. Replication techniques are widely used to increase the availability of data, improving performance of query latency and load balancing in Data Grid. Also effective resource scheduling is a challenging research issue. In this paper we propose a job scheduling policy, called Parallel Job Scheduling (PJS), and a dynamic data replication strategy, called Threshold-based Dynamic Data Replication (TDDR), to improve the data access efficiencies in a hierarchical Data Grid. The PJS uses hierarchical scheduling to reduce the search time for an appropriate computing node. It considers network characteristics, number of jobs waiting in queue, file locations, and disk read speed of storage drive at data sources. The main idea of TDDR strategy is using a threshold value to determine if the requested replica needs to be copied to the node. The TDDR determines this threshold dynamically based on data request arrival rates and available storage capacities. Then, in order to overcome the problem of limited storage space in each node, we design an efficient replica replacement strategy, which is developed as a two stages process. First, it deletes those files with minimum time for transferring. Second, if space is still insufficient then it considers the last time the replica was requested, number of access, size of replica and file transfer time. Results from the simulation show that our proposed algorithms have better performance in comparison with other algorithms in terms of Mean Job Time, Number of Intercommunications, Number of Replications, Computing Resource Usage, and Effective Network Usage.

Extending goal-oriented parallel computer job scheduling policies to heterogeneous systems

Several existing parallel computer systems partition their system resources or consist of systems from different geographical locations. This work is focusing on extending the original goal-oriented parallel computer job scheduling policies to cover such systems. The goal-oriented parallel computer job scheduling policies are proposed recently to handle conflicting objectives by utilizing a combinatorial search technique to find the most compromise schedule within a time limit. In this paper, some modifications to the original goal-oriented parallel computer job scheduling policy design are proposed and evaluated. The proposed policy is evaluated against basic priority backfilling techniques widely used in the field. Both homogeneous and heterogeneous parallel computer systems in terms of the computing power are evaluated. And, the design decision on the partition selection heuristic is also evaluated in this study. The experimental results show that the proposed policy produces good scheduling performances even when inaccurate runtime information is used.

An Adaptive Scheduler Framework for Complex Workflow Jobs on Grid Systems

Grid Computing provides sharing of geographically distributed resources among large scale complex applications. Due to dynamic nature of resources in grid, there is a need of highly efficient job scheduling and resource management policies in grid. A novel Grid Resource Scheduler (GRS) is proposed to effectively utilize the available resources in Grid. Proposed GRS contributes, an optimal job scheduling algorithm on Job Rank-Backfilling policy and a resource matching algorithm based on ranking of resources with best fit allocation model. Performance of GRS is measured by considering a web based BLAST algorithm – a bioinformatics application. GRS aims in reducing; Makespan of the job workflow, execution time of varied size jobs, response time of the submitted jobs and overhead of using the system. It also considers improving the utilization factor and throughput of the available heterogeneous resources in grid. The experimental results prove that proposed grid scheduler framework performs better when evaluated against widely used First Come First Serve (FCFS), Shortest Job First (SJF) and Minimum Time to Due Date (MTTD) scheduling strategies.

Parallel job scheduling for power constrained HPC systems

Power has become the primary constraint in high performance computing. Traditionally, parallel job scheduling policies have been designed to improve certain job performance metrics when scheduling parallel workloads on a system with a given number of processors. The available number of processors is not anymore the only limitation in parallel job scheduling. The recent increase in processor power consumption has resulted in a new limitation: the available power. Given constraints naturally lead to an optimization problem. We proposed MaxJobPerf, a new parallel job scheduling policy based on integer linear programming. Dynamic Voltage Frequency Scaling (DVFS) is a widely used technique that running applications at reduced CPU frequency/voltage trades increased execution time for power reduction. The optimization problem determines which jobs should run and at which frequency. In this paper, we compare the MaxJobPerf policy against other power budgeting policies for different power budgets. It clearly outperforms the other power-budgeting approaches at the parallel job scheduling level. Furthermore, we give a detailed analysis of the policy parameters including a discussion on how to manage job reservations to avoid job starvation.

Grid scheduling algorithm based on application diversity

Grid computing was initially proposed to harness huge amounts computational resources. In grid and parallel systems, backfilling has proven to be very efficient method when parallel jobs are considered. Backfilling is nonFCFS job scheduling policy that allows the jobs to move ahead in queue, given they don’t delay previously submitted jobs. We present an extension to the existing conservative backfill by creating multiple queues for jobs based on their runtime along with earliest deadline first (EDF) scheduling policy, which improves the performance of the algorithm in terms of Average weighted response time (AWRT) and average slowdown of the jobs.

Design of Stochastic Simulator for Analyzing the Impact of Scalability on CPU Scheduling Algorithms

Process scheduling with scalable performance is an issue in computer system. Scalability of scheduling algorithm is its ability to don’t decrease the performance when large processes are under run. The performance of job scheduling policies strongly depends on the properties of the incoming jobs. In this paper, we have analyzed the impact of scalability on different CPU scheduling algorithms with reference to average waiting time, average turnaround time and average response time to determine which algorithm is most suitable for uniprocessor environment. The burst time, arrival time and priority is randomly generated using exponential probability distribution and the performance of all algorithms has been evaluated with reference to arrival time or without arrival time. We use a simulative approach to evaluate the performance and scalability of each algorithm with reference to different number of processes.

Improvement of Job Scheduling and Tow Level Data Replication Strategies in Data Grid

With the development of grid computing, Data Grid, as an important branch of Grid Computing focuses on supporting an efficient management mechanism for controlled sharing and large amounts of distributed data. Data intensive jobs are one major kind of jobs in Data Grid, and their scheduling strategies are regarded as one of the most important research fields. Dynamic replication in data grid aims to reduce data access time and to utilize network and storage resources efficiently. This paper proposes a novel dynamic replication strategy, called ERS(Enhancement of Replication Strategy), which reduces data access time by avoiding network congestions in a data grid network and we develop a job scheduling policy, called ESS (Enhancement of Scheduling strategy).

The grid, the load and the gradient

An important concern for an efficient use of distributed computing is dealing with load balancing to ensure all available nodes and their shared resources are equally exploited. In large scale systems such as volunteer computing platforms and desktop grids, centralized solutions may introduce performance bottlenecks and single points of failure. Accordingly fully distributed alternatives have been considered, due to their inherent robustness and reliability. In extremely dynamic contexts, scheduling middlewares should adapt their job scheduling policies to the actual availability and overcome the volatility and heterogeneity typical of the underlying nodes. To deal with the dynamicity of a large pool of resources, self-organizing and adaptive solutions represent a promising research direction. Solutions based on bio-inspired methodologies are particularly suitable, as they inherently provide the desired features. In this paper we present a fully distributed load balancing mechanism, called ozmos, which aims at increasing the efficiency of distributed computing systems through peer-to-peer interaction between nodes. The proposed algorithm is based on a Chord overlay, and employs ant-like agents to spread information about the current load on each node, to reschedule tasks from overloaded systems to underloaded ones, and to relocate incompatible tasks on suitable resources in heterogeneous grids. By means of several evaluation scenarios we demonstrate the effectiveness of the proposed solution in achieving system-wide load balancing, both with homogeneous and heterogeneous resources. In particular we consider the load balancing performance of our approach, its scalability, as well as its communication efficiency.

English

In grid scheduling systems, a major challenge is to manage the consumers’ job based on their Quality of Service (QoS) and provider nodes’ satisfaction. Most of the capable job scheduling polices operate on the basis of meta-scheduling systems that may result in a complex management with overcrowding probability in market-oriented grids. Therefore, the need for an efficient job scheduling algorithm based on the local scheduling policy is vital to reduce overcrowding in the meta-scheduling system. This paper presents an efficient scheduling approach concerning the localization of job scheduling policy. This approach adapts appropriate computing nodes to jobs by taking their throughput into account and QoS requirements. The job management policy of the proposed scheduling approach focuses specially on increasing the job submission rate using accurate estimation procedure and completing the submitted jobs within defined deadline. Experiments were designed to study the performance of the proposed job scheduling approach. We compared the performance of the proposed approach with other popular algorithms and policies based on general and standard metrics. Results show an increase in the performance and user satisfaction criteria. Additionally, the overdue time for jobs which is a main concern in market-oriented grid computing systems was significantly improved. Key words: Hierarchical job scheduling, quality of service, cooperative-based scheduling, market-oriented grid.

On/off-line prediction applied to job scheduling on non-dedicated NOWs

This paper proposes a prediction engine designed for non-dedicated clusters, which is able to estimate the turnaround time for parallel applications, even in the presence of serial workload of the workstation owner. The prediction engine can be configured to work with three different estimation kernels: a Historical kernel, a Simulation kernel based on analytical models and an integration of both, named Hybrid kernel. These estimation proposals were integrated into a scheduling system, named CISNE, which can be executed in an on-line or off-line mode. The accuracy of the proposed estimation methods was evaluated in relation to different job scheduling policies in a real and a simulated cluster environment. In both environments, we observed that the Hybrid system gives the best results because it combines the ability of a simulation engine to capture the dynamism of a non-dedicated environment together with the accuracy of the historical methods to estimate the application runtime considering the state of the resources.

Two Level Job Scheduling and Data Replication in Data Grid

Data Grid environment is a geographically distributed that deal with date-intensive application in scientific and enterprise computing. In data-intensive applications data transfer is a primary cause of job execution delay. Data access time depends on bandwidth, especially when hierarchy of bandwidth appears in network. Effective job scheduling can reduce data transfer time by considering hierarchy of bandwidth and also dispatching a job to where the needed data are present. Additionally, replication of data from primary repositories to other locations can be an important optimization step to reduce the frequency of remote data access. Objective of dynamic replica strategies is reducing file access time which leads to reducing job runtime. In this paper we develop a job scheduling policy, called TLSS (Two level Scheduling strategy), and a dynamic data replication strategy, called TLRS (Two level Replication Strategy), to improve the data access efficiencies in a cluster grid. We study our approach and evaluate it through simulation. The results show that combination of TLSS and TLRS has improved 17% over other combinations.

The Impact of Data Replication on Job Scheduling Performance in Hierarchical Data Grid

In data-intensive applications data transfer is a primary cause of job execution delay. Data access time depends on bandwidth. The major bottleneck to supporting fast data access in Grids is the high latencies of Wide Area Networks and Internet. Effective scheduling can reduce the amount of data transferred across the internet by dispatching a job to where the needed data are present. Another solution is to use a data replication mechanism. Objective of dynamic replica strategies is reducing file access time which leads to reducing job runtime. In this paper we develop a job scheduling policy and a dynamic data replication strategy, called HRS (Hierarchical Replication Strategy), to improve the data access efficiencies. We study our approach and evaluate it through simulation. The results show that our algorithm has improved 12% over the current strategies.

Utilization driven power-aware parallel job scheduling

In this paper, we propose UPAS (Utilization driven Power-Aware parallel job Scheduler) assuming DVFS enabled clusters. A CPU frequency assignment algorithm is integrated into the well established EASY backfilling job scheduling policy. Running a job at lower frequency results in a reduction in its power dissipation and energy consumption, but introduces a penalty in its performance. Furthermore, performance of other jobs may be affected as their wait times can increase. For this reason, we propose to apply DVFS when system utilization is below a certain threshold, exploiting periods of low system activity. As the increase in run times due to frequency scaling can be seen as an increase in computational load, we have done an analysis of HPC system dimension. This paper investigates whether having more DVFS enabled processors and scheduling jobs with UPAS can lead to lower energy consumption and higher performance. Five workload traces from systems in production use with up to 9 216 processors are simulated to evaluate the proposed algorithm and the dimensioning problem. Our approach decreases CPU energy by 8% on average depending on allowed job performance penalty. Applying UPAS to 20% larger systems, CPU energy needed to execute same workloads can be decreased by 20% while having same or better job performance.

Energy-efficient scheduling for multiple access in wireless sensor networks: A job scheduling method

With the objective to minimize the transmission energy cost, we consider the energy-efficient scheduling problem in a single hop multi-access data gathering network. We first prove by theoretical induction that transmitting with reduced powers decreases the energy budget in a multi-access transmission. Given the optimal transmit powers, we then examine the multi-access capacity polymatroid and argue that the optimal rate control can be achieved by controlling the successive decoding order of the transmitting sensor nodes. Consequently, the multi-access scheduling problem is reformulated into the job scheduling problems and solved by adapting job scheduling policies. We also address the implementation issues and demonstrate via simulations that the proposed strategies are efficient in energy conservation.

Job scheduling and data replication on data grids

In data grids, many distributed scientific and engineering applications often require access to a large amount of data (terabytes or petabytes). Data access time depends on bandwidth, especially in a cluster grid. Network bandwidth within the same cluster is larger than across clusters. In a communication environment, the major bottleneck to supporting fast data access in Grids is the high latencies of Wide Area Networks (WANs) and Internet. Effective scheduling in such network architecture can reduce the amount of data transferred across the Internet by dispatching a job to where the needed data are present. Another solution is to use a data replication mechanism to generate multiple copies of the existing data to reduce access opportunities from a remote site. To utilize the above two concepts, in this paper we develop a job scheduling policy, called HCS (Hierarchical Cluster Scheduling), and a dynamic data replication strategy, called HRS (Hierarchical Replication Strategy), to improve the data access efficiencies in a cluster grid. We simulate our algorithm to evaluate various combinations of data access patterns. We also implement HCS and HRS in the Taiwan Unigrid environment. The simulation and experiment results show that HCS and HRS successfully reduces data access time and the amount of inter-cluster-communications in comparison with other strategies in a cluster grid.

A decentralized resource allocation policy in minigrid

Resource allocating is the key to enhance the concurrency ability of the heterogeneous distributed systems. The traditional resource allocating policies statically assign resources to the jobs according to the distribution schema computed by the job scheduling policy. Those policies cannot handle the DAGs with large jobs since the computing cost is intolerably huge. Decentralized resource allocating policy is inevitably generated to solve this problem by sharing the computing burden on several processors. However the existing decentralized policies cannot dynamically compute the dependent relationships of a given DAG or even schedule the data driven jobs. This paper proposes a cluster-based resource allocating policy—CDRAP, by which the large-scale distributed systems can precisely schedule the data driven jobs and the computing cost of jobs’ scheduling can be controlled in an acceptable range as well. The Broker + describe/issue message mechanism is used as the underlying structure in CDRAP.

Robust scheduling of moldable parallel jobs

Moldable job scheduling has been shown to be promising compared with traditional job scheduling policies. In contrast to rigid job scheduling where the number of processors for a job is fixed by the user, with moldable job scheduling the number of processors allocated to each job is chosen by the scheduler from a given range. Using simulation-based experiments, we show that previously proposed moldable scheduling strategies are not very robust overall performance improves in some circumstances, but worsens under other circumstances, and considerably degrades performance for some categories of jobs. This paper proposes a more robust strategy for scheduling moldable jobs, which outperforms the rigid scheduling scheme and previous moldable scheduling strategies. The new strategy performs well under different load conditions and for jobs of different scalabilities.

Minimizing cycle time and group scheduling, using Petri nets a study of heuristic methods

The objective of this paper is a study of minimizing the maximum completion time min F max, or cycle time of the last job of a given family of jobs using flow shop heuristic scheduling techniques. Three methods are presented: minimize idle time (MIT); Campbell, Dudek and Smith (CDS); and Palmer. An example problem with ten jobs and five machines is used to compare results of these methods. A deterministic t-timed colored Petri net model has been developed for scheduling problem. An execution of the deterministic timed Petri net allows to compute performance measures by applying graph traversing algorithms starting from initial global state and going into a desirable final state(s) of the production system. The objective of the job scheduling policy is minimizing the cycle time of the last job scheduled in the pipeline of a given family of jobs. Three heuristic scheduling methods have been implemented. First, a sub-optimal sequence of jobs to be scheduled is generated. Second, a Petri net-based simulator with graphical user interface to monitor execution of the sequence of tasks on machines is dynamically designed. A deterministic t-timed colored Petri net model has been developed and implemented for flexible manufacturing systems (FMS). An execution of the deterministic timed Petri net into a reachability graph allows to compute performance measures by applying graph traversing algorithms starting from initial global state to a desirable final state(s) of the production system.