Batch Jobs Research Articles

Column generation for minimizing total completion time on a single machine with parallel batching

Abstract In manufacturing of integrated circuits, burn in operations are key operations to improve the quality by detecting early failures. Such operations are performed in the so-called batch ovens, able to process several circuits at the same time. In this paper, the problem of deciding in which sequence to process a given number of circuit boards in a batch oven is addressed. This corresponds to a single-machine scheduling problem on a machine with parallel batch job processing. A column generation algorithm is developed for the minimization of the total completion time when circuits have an individual size. The computational results show the efficiency and effectiveness of the proposed algorithm, able to solve instances up to 100 jobs.

Distributed Join Processing Between Streaming and Stored Big Data Under the Micro-Batch Model

In order to interpret, enrich, and analyze the streaming data, stream applications often access the data stored in an external database. Although there has been a lot of studies on stream processing, little attention has been paid so far to the join between streaming data and stored data. In this paper, we propose a comprehensive solution called DS-join for distributed processing of the join under the micro-batch model of recently distributed stream processing engines (SPEs), such as spark streaming. The micro-batch model performs stream processing as a series of very small batch jobs and is more fault-tolerant in a distributed environment compared with the record-at-a-time model. The DS-join reduces the number of database accesses by using micro-batching. Furthermore, the DS-join optimizes the join operation by minimizing the data shuffling, managing a cache in a distributed SPE, parallelizing the join processing, and balancing the load between the SPE and the external database system. The experimental results using real and synthetic datasets show that, compared with the state-of-the-art methods, the DS-join significantly improves throughput, especially for large databases.

Minimize the Make-span of Batched Requests for FPGA Pooling in Cloud Computing

Using FPGA as accelerators is gaining popularity in Cloud computing. Usually, FPGA accelerators in a datacenter are managed as a single resource pool. By issuing a request to this pool, a tenant can transparently access FPGA resources. FPGA requests usually arrive in batches. The objective of scheduling is to minimize the make-span of a given batch of requests, which is the completion time of the entire batch of jobs. As a result, either the responsiveness is improved, or the system throughput is maximized. The key technical challenge is the existence of multiple resource bottlenecks. An FPGA job can be bottlenecked by either computation (i.e., computation-intensive) or network (i.e., network-intensive), and sometimes by both. To the best of our knowledge, this is the first work that minimizes the make-span of batched requests for an FPGA accelerator pool in Cloud computing that considers multiple resource bottlenecks. In this paper, we design several scheduling algorithms to address the challenge. We implement our scheduling algorithms in an IBM Cloud system. We conduct extensive evaluations on both a small scale testbed and a large-scale simulator. Compared with the Shortest-Job-First scheduling, our algorithms can reduce the make-span by 36.25 percent, and improve the system throughput by 36.05 percent.

Adaptive Threshold Based Scheduler for Batch of Independent Jobs for Cloud Computing System

Distributed systems are efficient means of realizing high-performance computing (HPC). They are used in meeting the demand of executing large-scale high-performance computational jobs. Scheduling the tasks on such computational resources is one of the prime concerns in the heterogeneous distributed systems. Scheduling jobs on distributed systems are NP-complete in nature. Scheduling requires either heuristic or metaheuristic approach for sub-optimal but acceptable solutions. An adaptive threshold-based scheduler is one such heuristic approach. This work proposes adaptive threshold-based scheduler for batch of independent jobs (ATSBIJ) with the objective of optimizing the makespan of the jobs submitted for execution on cloud computing systems. ATSBIJ exploits the features of interval estimation for calculating the threshold values for generation of efficient schedule of the batch. Simulation studies on CloudSim ensures that the ATSBIJ approach works effectively for real life scenario.

Stochastic Modeling and Optimization of Stragglers

MapReduce framework is widely used to parallelize batch jobs since it exploits a high degree of multi-tasking to process them. However, it has been observed that when the number of servers increases, the map phase can take much longer than expected. This paper analytically shows that the stochastic behavior of the servers has a negative effect on the completion time of a MapReduce job, and continuously increasing the number of servers without accurate scheduling can degrade the overall performance. We analytically model the map phase in terms of hardware, system, and application parameters to capture the effects of stragglers on the performance. Mean sojourn time (MST), the time needed to sync the completed tasks at a reducer, is introduced as a performance metric and mathematically formulated. Following that, we stochastically investigate the optimal task scheduling which leads to an equilibrium property in a datacenter with different types of servers. Our experimental results show the performance of the different types of schedulers targeting MapReduce applications. We also show that, in the case of mixed deterministic and stochastic schedulers, there is an optimal scheduler that can always achieve the lowest MST.

PyCosmo: An integrated cosmological Boltzmann solver

Abstract As wide-field surveys yield ever more precise measurements, cosmology has entered a phase of high precision requiring highly accurate and fast theoretical predictions. At the heart of most cosmological model predictions is a numerical solution of the Einstein–Boltzmann equations governing the evolution of linear perturbations in the Universe. We present PyCosmo , a new Python-based framework to solve this set of equations using a special purpose solver based on symbolic manipulations, automatic generation of C++ code and sparsity optimisation. The code uses a consistency relation of the field equations to adapt the time step and does not rely on physical approximations for speed-up. After reviewing the system of first-order linear homogeneous differential equations to be solved, we describe the numerical scheme implemented in PyCosmo . We then compare the predictions and performance of the code for the computation of the transfer functions of cosmological perturbations and compare it to existing cosmological Boltzmann codes. While PyCosmo does not yet have all the features of other codes, our approach is complementary to other fast cosmological Boltzmann solvers and can be used as an independent test of their numerical solutions. The symbolic representation of the Einstein–Boltzmann equation system in PyCosmo provides a convenient interface for implementing extended cosmological models. We also discuss how the PyCosmo framework can also be used as a general framework to compute cosmological quantities as well as observables for both interactive and high-performance batch jobs applications. Information about the PyCosmo package and future code releases are available at http://www.cosmology.ethz.ch/research/software-lab.html .

Special Fixture Design and Precision Measurement of Automobile Flange Fork

Because of structural features and batch job requirements in automotive manufacturing line of automotive flange fork parts, the high requirement about its parts is made in the aspect of four positioning datum hole position degrees. Therefore, it is necessary to consider special fixture when designing and processing its parts to make sure batch production, and precision measurement to ensure product quality. In the actual processing, three generations of special fixtures for processing and production use are designed to make improvement for processing and contrasting, through the process of design, trial processing, size adjustment, fixture precision measure and adjustment. Special fixture size precision adjustment is completed, through modern precision measurement equipment using three-dimensional coordinate measurement, to satisfy the parts processing quality and requirements. At present, this production technology has reference significance to a certain extent in the manufacturing process of automotive fork similar products.

Managing Risk in a Derivative IaaS Cloud

Infrastructure-as-a-Service (IaaS) cloud platforms rent computing resources with different cost and availability tradeoffs. For example, users may acquire virtual machines (VMs) in the spot market that are cheap, but can be unilaterally terminated by the cloud operator. Because of this revocation risk, spot servers have been conventionally used for delay and risk tolerant batch jobs. In this paper, we develop risk mitigation policies which allow even interactive applications to run on spot servers. Our System, SpotCheck is a derivative cloud platform, and provides the illusion of an IaaS platform that offers always-available VMs on demand for a cost near that of spot servers, and supports unmodified applications. SpotCheck’s design combines virtualization-based mechanisms for fault-tolerance, and bidding and server selection policies for managing the risk and cost. We implement SpotCheck on EC2 and show that it i) provides nested VMs with 99.9989 percent availability, ii) achieves upto 2-5 $\times$ cost savings compared to using on-demand VMs, and iii) eliminates any risk of losing VM state.

Investigation on Processing of Real-Time Streaming Big Data

MapReduce is the most widely used for huge data processing and it is a part of the Hadoop big data and this will provide the quality and efficient results because of their processing functions. For the batch jobs, Hadoop is the proper and also there is inflated request for non-batch elements homogeneous interactive jobs, and high data currents. For this non-batch assignments, consider Hadoop is not useful and present situations are recommending to these new crises. In this paper, these are divided into two stages that are real-time processing, and stream processing of big data. For every stage, the models are deliberate, stability and diversity to Hadoop. For every group, we have provided the working systems and structures. For the creation of the new examples, some experiments are conducted to improve the new results belongs to available Hadoop-based solutions.

Sojourn time in a Processor Sharing queue with batch arrivals

For the Processor Sharing queue with batch job arrivals, the sojourn time W of a single job is investigated. This queuing model is motivated by the evaluation of Cloud Computing or Virtualized Network systems where the treatment of micro-services within requests determines the global system performance. We first show that the distribution of sojourn time W can be obtained from an infinite linear differential system; the structure of this system, however, makes the explicit derivation of this distribution generally difficult. When further assuming that the batch size has a geometric distribution with some given parameter , this differential system can be analyzed via a single generating function which is shown to verify a second-order partial differential equation involving a boundary term at point . Solving this partial differential equation for with required analyticity properties determines the one-sided Laplace transform for given u. Writing in terms of a multivariate hypergeometric function enables one to extend its analyticity domain to a cut-plane , for negative constants and . By means of a Laplace inversion of for a suitable value of u, the complementary distribution function is then given an explicit integral representation. This enables us to show that the tail of this distribution has an exponential decay with rate , together with a sub-exponential factor. A convergence in distribution is also asserted for W in heavy load condition, the limit distribution exhibiting a sub-exponential behavior itself. Using our exact results for the sojourn time of a single job, we finally discuss an approximation for the distribution of the sojourn time of an entire batch when assuming that the batch size is not too large.

Exploiting Spatio-Temporal Diversity for Water Saving in Geo-Distributed Data Centers

As the critical infrastructure for supporting Internet and cloud computing services, massive geo-distributed data centers are notorious for their huge electricity appetites and carbon footprints. Nonetheless, a lesser-known fact is that data centers are also “thirsty”: to operate data centers, millions of gallons of water are required for cooling and electricity production. The existing water-saving techniques primarily focus on improved “engineering” (e.g., upgrading to air economizer cooling, diverting recycled/sea water instead of potable water) and do not apply to all data centers due to high upfront capital costs and/or location restrictions. In this paper, we propose a software-based approach towards water conservation by exploiting the inherent spatio-temporal diversity of water efficiency across geo-distributed data centers. Specifically, we propose a batch job scheduling algorithm, called WACE (minimization of WAter, Carbon and Electricity cost), which dynamically adjusts geographic load balancing and resource provisioning to minimize the water consumption along with carbon emission and electricity cost while satisfying average delay performance requirement. WACE can be implemented online without foreseeing the far future information and yields a total cost (incorporating electricity cost, water consumption and carbon emission) that is provably close to the optimal algorithm with lookahead information. Finally, we validate WACE through a trace-based simulation study and show that WACE outperforms state-of-the-art benchmarks: $25$ percent water saving while incurring an acceptable delay increase. We also extend WACE to joint scheduling of batch workloads and delay-sensitive interactive workloads for further water footprint reduction in geo-distributed data centers.

Priority scheduling with consolidation based backfilling algorithm in cloud

In cloud computing, scheduling plays an eminent role while processing enormous jobs. The paralle jobs utmost need parallel processing capabilities which leads to CPU underutilization mainly due to synchronization and communication among parallel processes. Researchers introduced several algorithms for scheduleing parallel jobs namely, Conservative Migration Consolidation supported Backfilling (CMCBF) and Aggressive Migration Consolidation supported Backfilling (AMCBF). The greatest challenge of a existing scheduling algorithm is to improve the data center utilization without affecting job responsiveness. Hence, this work proposes an Effective Multiphase Scheduling Approach (EMSA) to process the jobs. In EMSA, the jobs are initially preprocessed and batched together to avoid starvation and to mitigate unwanted delay. Later, an Associate Priority Method has been proposed which prioritizes the batch jobs to minimize the number of migrations. Finally, the prioritized jobs are scheduled using Priority Scheduling with BackFilling algorithm to utilize the intermediate idle nodes. Moreover, the virtualization technology partitions the computing capacity of the Virtual Machine (VM) into two-tier VM as foreground VM (FVM) and Background VM (BVM) to improve node utilization. Hence, Priority Scheduling with Consolidation based BackFilling algorithm has been deployed in a two-tier VM that processes the jobs by utilizing the VMs effectively. Experimental results show that the performance of the proposed work performs better than other existing algorithms by increasing the resource utilization by 8%.

An optimization framework for the capacity allocation and admission control of MapReduce jobs in cloud systems

Nowadays, we live in a Big Data world and many sectors of our economy are guided by data-driven decision processes. Big Data and Business Intelligence applications are facilitated by the MapReduce programming model, while, at infrastructural layer, cloud computing provides flexible and cost-effective solutions to provide on-demand large clusters. Capacity allocation in such systems, meant as the problem of providing computational power to support concurrent MapReduce applications in a cost-effective fashion, represents a challenge of paramount importance. In this paper we lay the foundation for a solution implementing admission control and capacity allocation for MapReduce jobs with a priori deadline guarantees. In particular, shared Hadoop 2.x clusters supporting batch and/or interactive jobs are targeted. We formulate a linear programming model able to minimize cloud resources costs and rejection penalties for the execution of jobs belonging to multiple classes with deadline guarantees. Scalability analyses demonstrated that the proposed method is able to determine the global optimal solution of the linear problem for systems including up to 10,000 classes in less than 1 s.

Scheduling step-deteriorating jobs on bounded parallel-batching machines to maximise the total net revenue

This paper addresses a parallel-batching scheduling problem considering processing cost and revenue, with the objective of maximising the total net revenue. Specifically, the actual processing time of a job is assumed to be a step function of its starting time and the common due date. This problem involves assigning jobs to different machines, batching jobs, and sequencing batches on each machine. Some key structural properties are proposed for the scheduling problem, based on which an optimal scheduling scheme is developed for any given machine. Then, an effective hybrid VNS–IRG algorithm which combines Variable Neighborhood Search (VNS) and Iterated Reference Greedy algorithm (IRG) is proposed to solve this problem. Finally, the effectiveness and stability of the proposed VNS–IRG are demonstrated and compared with VNS, IRG, and Particle Swarm Optimization through computational experiments.

Green IT scheduling for data center powered with renewable energy

In recent years, the question of the energy consumption of data centers has become more and more important, and several studies raised the possibility of using renewable energy to power them. The intermittent nature of commonly used renewable energy sources is a major drawback of using them directly on-site. In this paper, we present an approach for scheduling batch jobs with due date constraints, which takes into account the availability of the renewable energy to reduce the need of brown energy and therefore running cost. The approach we propose differs from the existing methods by providing a scheduling algorithm agnostic of the electrical infrastructure. A separated system, managing the renewable sources, provides an arbitrary objective function, which is used to guide the scheduling heuristic. We implemented our approach in a data center simulator, and evaluated it by considering a small-scale center powered with solar panels and connected to the electrical grid. The relationship between the flexibility allowed by the user negotiated SLAs and the behavior of the algorithm is studied, and compared to existing approaches from the literature. Our experiments show a reduction of brown energy consumption up to 49% and a cost saving up to 51%, compared to a traditional scheduler unaware of renewable availability.

Going Fast and Fair: Latency Optimization for Cloud-Based Service Chains

State-of-the-art microservices have been attracting more attention in recent years. A broad spectrum of online interactive applications are now programmed to service chains on the cloud, seeking better system scalability and lower operating costs. Different from the conventional batch jobs, most of these applications consist of multiple stand-alone services that communicate with each other. These step-by-step operations unavoidably introduce higher latency to the delay-sensitive chained services. In this article, we aim at designing an optimization approach for reducing the latency of chained services. Specifically, presenting the measurement and analysis of chained services on Baidu's cloud platform, our real-world trace indicates that these chained services are suffering from significantly high latency because they are mostly handled by different queues on cloud servers for multiple times. However, such a unique feature introduces significant challenges to optimize a microservice's overall queueing delay. To address this problem, we propose a delay-guaranteed approach to accelerate the overall queueing of chained services while obtaining fairness across all the workloads. Our evaluations on Baidu servers shows that the proposed design can successfully reduce the latency of chained services by 35 percent with minimal impact on other workloads.

병렬처리 기법을 이용한 복합재 적층 구조물의 면내 파손 해석 소프트웨어 개발

본 논문에서는 복합재 적층 구조물의 면내 구조 해석을 위한 자동화 소프트웨어를 개발한다. 개발된 소프트웨어는 다양한 파손 기준을 제공하며 사용자의 편리성을 고려하여 해석 결과를 출력한다. 또한 병렬처리 기법에 기반한 일괄 처리 해석(batch job analysis) 기능을 지원한다. 구현된 소프트웨어의 성능을 검증하기 위하여 소프트웨어에서 계산된 안전 여유와 자체적으로 사용 중인 in-house 방식과 실제 시편 실험에서 얻어진 안전 여유간의 비교를 수행하였다. 비교 결과 in-house 방식과는 0.01 이하의 오차가 있었으며 시편 실험과는 약 ±10% 이내의 오차가 존재함을 확인하였다. 또한 병렬처리 기법을 적용한 일괄 처리 해석 작업의 실행 속도 개선을 확인하였다.

Modelling and symmetry breaking in scheduling problems on batch processing machines

Problems of scheduling batch-processing machines to minimise the makespan are widely exploited in the literature, mainly motivated by real-world applications, such as burn-in tests in the semiconductor industry. These problems consist of grouping jobs in batches and scheduling them on machines. We consider problems where jobs have non-identical sizes and processing times, and the total size of each batch cannot exceed the machine capacity. The processing time of a batch is defined as the longest processing time among all jobs assigned to it. Jobs can also have non-identical release times, and in this case, a batch can only be processed when all jobs assigned to it are available. This paper discusses four different versions of batch scheduling problems, considering a single processing machine or parallel processing machines and considering jobs with or without release times. New mixed integer linear programming formulations are proposed as enhancements of formulations proposed in the literature, and symmetry breaking constraints are investigated to reduce the size of the feasible sets. Computational results show that the proposed formulations have a better performance than other models in the literature, being able to solve to optimality instances only considered before to be solved by heuristic procedures.

TOWARDS THE LEVELING OF MULTI-PRODUCT BATCH PRODUCTION FLOWS. A MULTIMODAL NETWORKS PERSPECTIVE.

The problem studied in this paper is a cyclic job-shop problem with multiple Automated Guided Vehicles (AGVs). Job batches, which follow specific production routes, are processed, in the order of their operations, on multiple machines with standard processing times, and the fleet of AGVs perform the transportation operations of moving these job batches between the workstations. In this system, part sets of items are produced at fixed time intervals (takts). In the adopted model of the system, one can distinguish a layer of station-to-station transport, which is a network of local loops connecting subsets of workstations serviced cyclically by dedicated AGVs, and a layer of repetitive production flows which comprise job batches following a given set of production routes. The relationship between the elements of the structure of the system being modelled and its potential behavior is described by a system of integer equations. The resulting solutions enable the fast evaluation of production flow parameters including part sets, takt time, as well as repetitive-flow balancing aimed at maximization of the rate of system resource utilization. The high efficiency of the proposed approach, enabling the online prototyping of the production flow, is a consequence of omitting the time-consuming calculation of the sequencings of jobs within a cycle.

A Green Mathematical Model for a Single-Machine Scheduling Problem with Batch Delivery System

This paper aims to minimize the total tardiness and the sum of delivery costs of a single machine problem when jobs are delivered to customers in various size batches. In the real world, this issue may happen within a supply chain in which submitting jobs to customers entails costs. On the one hand, to avoid delay costs manufacturers should deliver jobs to customers immediately after their completion, which leads to more number of transportation. Consequently, delivery costs related to the cost of vehicles traffic and their CO2 gas emissions increase. On the other hand, manufacturers should submit jobs in batches to reduce delivery costs and make the green supply chain. However, dispatching several jobs in the form of batches can have a negative effect on other scheduling-related objective functions such as minimizing the total tardiness. We present a mixed-integer linear programming (MILP) model to provide a trade-off between the total tardiness and the total delivery cost. As this is an MILP model, the commercial solver (CPLEX) is not guaranteed to find the optimal solution for large-size problems in a reasonable amount of time. Therefore, we present a genetic algorithm to solve the problem.