Task Placement Research Articles

Integer Programming Models for Task Scheduling and Resource Allocation in Mobile Cloud Computing

In traditional mobile cloud computing, user tasks are uploaded and processed on a cloud server over the Internet. Due to the recent rapid increase in the number of mobile users connected to the network, due to overload of the Internet communication channels, there are significant delays in the delivery of data processed on cloud servers to the user. Furthermore, it complicates the optimal scheduling of the tasks of many users on cloud servers and the delivery of results. Scheduling is an approach used to reduce the tasks execution time by ensuring a balanced distribution of user tasks on cloud servers. The goal of scheduling is to ensure selection of appropriate resources to handle tasks quickly, taking into account user requirements. Whereas the goal of cloud service providers is to provide users with the required resources through performing effective scheduling so that both the user and the service provider can benefit. The article proposes a scheduling model to reduce processing time, network latency, and power consumption of mobile devices through optimal task placement in the cloudlet network in a mobile cloud computing environment.

PipePar: Enabling fast DNN pipeline parallel training in heterogeneous GPU clusters

Recently, pipeline parallelism for large-scale Deep Neural Network (DNN) training has been developed, which partitions the DNN model across multiple devices (e.g., GPUs) and improves the training efficiency by processing data divided into minibatches as a pipeline. However, existing model partitioning algorithms are mostly designed for homogeneous clusters with the same GPU devices and network connections (e.g., bandwidths), while heterogeneous GPU clusters are widely used in mainstream computing infrastructures. In heterogeneous environment, devices are equipped with different GPUs and network connections, and the efficiency of previous approaches is unsatisfactory due to the unbalanced load of the pipeline stages. In this paper, we propose PipePar, a model partitioning and task placement algorithm for pipeline parallel DNN training in heterogeneous GPU clusters. PipePar is based on dynamic programming with search space pruning that takes into consideration both the heterogeneity of GPUs and network bandwidth. PipePar can profile the DNN model for each type of GPU and conduct model partitioning and task placement based on given GPUs and network connections, which can optimize pipeline load balancing in heterogeneous environments and thus improve training efficiency. We design and implement a pipeline-based distributed deep learning training system in a heterogeneous GPU cluster and show through extensive experiments that PipePar outperforms the baseline approaches in the speed of large-scale DNN training.

Eliminating Control Plane Overload via Measurement Task Placement

Recent efforts in network measurement place measurement tasks on programmable switches to measure high-speed traffic. These tasks extract flow data, i.e., events, from packets and send events to the control plane. However, the tasks may generate massive events in a short time. In this context, the links transferring events to the control plane and the control plane servers that handle events may be overloaded, i.e., control plane overload. None of existing solutions can eliminate control plane overload. In this paper, we propose, a framework that eliminates control plane overload via careful measurement task placement. Our key idea is to allocate enough resources for each task during task placement to avoid control plane overload at runtime. For each task, estimates its maximum possible rate of sending events to the control plane. Then its optimization framework addresses the resource restrictions of both switches and the control plane. The experiments on Tofino switches indicate that outperforms existing solutions with higher accuracy in several use cases.

Cloud service analysis using round-robin algorithm for quality-of-service aware task placement for internet of things services

Round-robin (RR) is a process approach to sharing resources that requires each user to get a turn using them in an agreed order in cloud computing. It is suited for time-sharing systems since it automatically reduces the problem of priority inversion, which are low-priority tasks delayed. The time quantum is limited, and only a one-time quantum process is allowed in round-robin scheduling. The objective of this research is to improve the functionality of the current RR method for scheduling actions in the cloud by lowering the average waiting, turnaround, and response time. CloudAnalyst tool was used to enhance the RR technique by changing the parameter value in optimizing the high accuracy and low cost. The result presents the achieved overall min and max response times are 36.69 and 650.30 ms for running 300 min RR. The cost for the virtual machines (VMs) is identified from $0.5 to $3. The longer the time used, the higher the cost of the data transfer. This research is significant in improving communication and the quality of relationships within groups.

ОРГАНИЗАЦИОННО-МЕТОДИЧЕСКИЕ ПОДХОДЫ К ИЗУЧЕНИЮ МЛАДЕНЧЕСКОЙ СМЕРТНОСТИ

The review reflects the relevance of the problem of infant mortality in the Russian Federation. Ap-proaches to the study of infant mortality were analyzed. The main risk factors for congenital patholo-gy of children, its conditioning by the presence of extragenital pathology of female mothers are pre-sented. The results of regional studies showing both the dynamics of infant mortality and its structure are reflected. The design approaches of infant mortality management, the latest technological devel-opments, achievements in basic and applied medical science are analyzed. The need to use task placement models in the maternity and childhood service aimed at timely diagnosis and preservation of the health of the child population has been demonstrated. The application of screening programs for the presence of congenital defects aimed at their early detection, optimal treatment and rehabilita-tion was assessed. The need to predict the formation of congenital fetal malformations in pregnant women in the first trimester of pregnancy is shown.

Nimbus: Towards Latency-Energy Efficient Task Offloading for AR Services

Widespread adoption of mobile augmented reality (AR) and virtual reality (VR) applications depends on their smoothness and immersiveness. Modern AR applications applying computationally intensive computer vision algorithms can burden today's mobile devices, and cause high energy consumption and/or poor performance. To tackle this challenge, it is possible to offload part of the computation to nearby devices at the edge. However, this calls for smart task placement strategies in order to efficiently use the resources of the edge infrastructure. In this paper, we introduce Nimbus --- a task placement and offloading solution for a multi-tier, edge-cloud infrastructure where deep learning tasks are extracted from the AR application pipeline and offloaded to nearby GPU-powered edge devices. Our aim is to minimize the latency experienced by end-users and the energy costs on mobile devices. Our multifaceted evaluation, based on benchmarked performance of AR tasks, shows the efficacy of our solution. Overall, Nimbus reduces the task latency by ~4x and the energy consumption by ~77% for real-time object detection in AR applications. We also benchmark three variants of our offloading algorithm, disclosing the trade-off of centralized versus distributed execution.

Collaborative Content Caching and Task Offloading in Multi-Access Edge Computing

Mobile augmented reality (MAR) applications are gaining popularity. There are significant challenges in handling the latency-sensitive and computation-intensive tasks generated by MAR applications on mobile devices. The emergence of mobile edge computing (MEC) provides a new idea to improve the computing capability of resource-constrained mobile terminals. In this paper, we study the task offloading and cache placement of MAR tasks in multi-MEC server cooperation system. The problem of the task offloading and cache placement is formulated to maximize the hit ratio and minimize the service latency under the constraints of MEC server's computing resources and cache space. To solve this problem, we propose a task offloading and cache placement (MOTOCP) algorithm based on multi-objective artificial bee colony. Pareto optimal relation is introduced in the optimization process to find the optimal solution. Extensive evaluation verifies that our proposed algorithm has better performance.

에너지 소모량 최소화를 위한 가상머신과 사용자 태스크 배치 알고리즘

클라우드에서 가상 머신을 최적의 물리 머신에 배치하는 전략과 사용자 태스크를 최적의 가상 머신에 배치하는 전략은 데이터 센터의 효율적인 전력 관리에 상당한 영향을 미치기 때문에 중요하다. 기존 연구는 데이터센터의 전체 가동 중인 물리 머신의 개수를 최소화하도록 가상머신과 사용자 태스크를 물리 머신에 배치함으로써 낭비되는 유휴 전력을 줄였다. 하지만 기존 연구의 에너지 소모량 최적화 모델은 물리 머신의 가동시간이 아닌 특정 임의의 시간만큼 계산하기 때문에 정확도가 낮다. 또한, 최적화 알고리즘인 Best fit은 지역최적해에 수렴한다는 단점이 존재한다. 본 논문에서는 기존 연구보다 더 높은 정확도의 에너지 소모량 모델을 제안하였고, 지역 최적해 수렴의 위험에서 벗어나는 유전자 알고리즘을 수정한 최적화 알고리즘을 제안하였다. 실험은 CloudSim 시뮬레이터를 통해서 제안 방안과 기존 논문의 방안에 대해서 성능을 비교 평가하였다. 실험 결과 본 논문의 제안 방안이 기존 논문의 방안보다 에너지를 더 효율적으로 소모함을 확인하였다.

Analisis Penempatan Tugas dan Fungsi Kerja Tenaga Kesehatan di UPTD Puskesmas Binjai Kota

Health Human Resources in health services is a form of efforts to plan, train, and develop health workers to support the achievement of the highest degree of health. The availability of quality health human resources, distributed fairly and evenly can ensure the implementation of health development. Puskesmas became the first health service facility for the community with promotive and preventive efforts. The lack of manpower and the uneven distribution of health workers in puskesmas will affect the degree of public health. This study aims to see how the placement of duties and work functions of health workers in UPTD Puskesmas Binjai Kota based on aspects of procurement, development, and maintenance. The type of research used is qualitative research with a descriptive approach and literature studies, namely observation, in-depth interviews and documentation. The results of the study obtained that the UPTD Puskesmas Binjai Kota has an excessive number of health workers with task placements that are not in accordance with their expertise and the low capacity of their human resources for computer operations. Therefore, proper planning and self-awareness of each individual are needed so that services at puskesmas can run optimally.

VioLinn: Proximity-aware Edge Placementwith Dynamic and Elastic Resource Provisioning

Deciding where to handle services and tasks, as well as provisioning an adequate amount of computing resources for this handling, is a main challenge of edge computing systems. Moreover, latency-sensitive services constrain the type and location of edge devices that can provide the needed resources. When available resources are scarce there is a possibility that some resource allocation requests are denied. In this work, we propose the VioLinn system to tackle the joint problems of task placement, service placement, and edge device provisioning. Dealing with latency-sensitive services is achieved through proximity-aware algorithms that ensure the tasks are handled close to the end-user. Moreover, the concept of spare edge device is introduced to handle sudden load variations in time and space without having to continuously overprovision. Several spare device selection algorithms are proposed with different cost/performance tradeoffs. Evaluations are performed both in a Kubernetes-based testbed and using simulations and show the benefit of using spare devices for handling localized load spikes with higher quality of service (QoS) and lower computing resource usage. The study of the different algorithms shows that it is possible to achieve this increase in QoS with different tradeoffs against cost and performance.

The Basic Elements of Devices Resource Consumption Decreasing Metodology for Distributed Systems on the Basis of Fog- and Edge-Computing

Purpose of research. The purpose of this study is to form a set of basic elements of the methodology for reducing the consumption of the residual resource of computing devices operating as part of distributed computing systems based on the concepts of fog and edge computing. The concepts of fog and edge computing are relatively new and, despite the large volume of publications on this topic, the issue of resource consumption of computing devices in terms of FBG values has not been considered in the literature. At the same time, extending the service life of devices is currently highly desirable, which makes this study relevant.Methods. The main scientific methods used in this study are analysis (of subject areas), numerical simulation and natural experiment, confirming the feasibility of the main aspects of the developed methodology.Within the framework of the concepts of fog and edge computing, it is considered appropriate to shift the computing load to data sources, which, as a rule, are located at the edge of the network. However, modern studies do not affect the estimates of the impact of such a strategy in the placement of functional tasks on the estimated values of the probability of non-failure operation of devices, which characterizes the state of the residual resource of the device. Meanwhile, an increase in the load on devices with less computing power than, say, a device within a data center leads to an acceleration of their wear, which, in turn, translates into economic costs for maintaining a functioning computing infrastructure. At the same time, the load on the intermediate network devices is reduced, since they transmit reduced amounts of data, and the time that can be used for data processing, if the latter is performed at the edge devices, increases. The developed methodology offers an integrated approach to the placement of functional tasks of distributed information systems, taking into account the listed features of using the concepts of fog and edge computing.Results. The main results of this study are the description of a set of basic methods that make up the methodology for reducing the consumption of the residual resource of computing devices of distributed computing systems based on fog and edge computing. The resulting complex is based on the developed models and the results of experimental studies.Conclusion. Currently, despite the massive use of the concepts of fog and edge computing in the implementation of distributed information systems, there has not been developed a unified methodology that would reduce the consumption of resources of computing devices and thereby extend their service life. Within the framework of this work, a set of methods is proposed, the further development of which will increase the service life of devices that make up the computing infrastructure of distributed computing systems.

An intelligent scheduling framework for DNN task acceleration in heterogeneous edge networks

With the upgrade of hardware architecture and device capacities, many accelerator-based hardware platforms have been widely deployed in Mobile Edge Computing (MEC) environments. The execution time of many computation-intensive applications (e.g., face recognition and pedestrian detection) can be significantly reduced when deployed on these heterogeneous devices. Moreover, thanks to the popularity of Deep Learning (DL), most terminal applications are integrated with Deep Neural Networks (DNN) and can be divided into interdependent tasks. The structure of these applications can be represented as the Directed Acyclic Graph (DAG). Therefore, it is critical to seek the optimal scheduling order and execution placement of tasks according to the acceleration effects of edge servers and the task dependency. However, conventional scheduling strategies focus on the short-term performance, potentially leading to service quality degradation in the long term. Besides, many studies use Deep Reinforcement Learning (DRL) algorithms to seek a long-term optimal scheduling strategy but ignore the device acceleration and the task dependency. Furthermore, training a well-performed DRL agent is time-consuming, and the large scale of trial-and-error will take up tremendous computation and storage resources. In this paper, we model the scheduling process as a Markov Decision Process (MDP) and design an adaptive scheduling framework for task acceleration. Fully considering the data dependencies, resource conditions, and network conditions, the proposed scheduling algorithm called Meta-AC uses policy gradient combined with meta-learning to minimize the average task delay and the ratio of time-out tasks. As a hierarchical DRL algorithm, Meta-AC uses meta data to learn directed exploration strategies in the high-level agent, improving the learning efficiency from experience samples. Extensive simulations demonstrate the superiority of the proposed method over the counterpart methods.

Efficient job scheduling paradigm based on hybrid sparrow search algorithm and differential evolution optimization for heterogeneous cloud computing platforms

The job scheduling paradigms include dispatching Internet of Things (IoT) critical services onto processing nodes. Here most energy is consumed in finding suitable virtual machines (VMs) that can execute IoT tasks without resource fragments. Therefore, a significant problem is minimizing energy consumption through efficient task placement that leads to load balance and minimizes resource leakage. To resolve this problem, we proposed a dual-phase metaheuristic algorithm called CSSA-DE. First, we conduct a clustering approach to group computing nodes into effective clusters. Each node is trained at different utilization levels, and the one that can yield the highest Performance-to-Power Ratio (PPR) is selected as the mega cluster head (MCH). Then, we integrated the sparrow search algorithm (SSA) with the differential evolution (DE) algorithm to expand the high search efficiency of finding an appropriate pair task-VM combination. Further, the integration phase can exploit the count of overloaded and underloaded VMs, reducing resource fragments. The performance of CSSA-DE is highly competitive and relatively better in multiple cases compared to state-of-the-art algorithms.

Synergies between resource sustainability and energy performance of cloud servers: The role of virtual machine repacking approach

Dynamic virtual machine (VM) consolidation to combine the workload of active devices and switch idle nodes into low-power states while preserving the quality of service (QoS) guarantee is the primary proposed solution’s goal to maintain dependable services between providers and customers. We propose a multi-step resource provision algorithm to address this issue efficiently. The forwarding step unites the iterative search for critical paths and the priority selection for task layers to reduce the makespan and reach an optimal critical task placement. The backward step trains processing servers against different loads using eclectic utilization levels to guarantee those servers function at their most efficient utilization values. Furthermore, this step optimizes energy consumption and resource diffusion by reallocating the existent VMs and mitigating an attempted overload of machines. Different evaluation metrics are assessed with respect to the number of tasks and VMs. The results show that the proposed paradigm optimizes energy consumption and resource utilization efficiently, lowering the makespan, overhead time, and network utilization and maximizing the throughput productivity and QoS satisfaction.

In-network Placement of Reusable Computing Tasks in an SDN-based Network Edge

Edge computing is aimed to support compute-intensive data-hungry interactive applications which can hardly run on resource-constrained consumer devices and may suffer from running in the cloud due to the long data transfer delay. The edge network nodes' heterogeneous and limited (compared to the cloud) capabilities make the computing task placement a challenge. In this paper, we propose a novel <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-network</i> task placement strategy aimed at minimizing the edge network resources usage. The proposal specifically accounts for <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">time-limited reusable</i> computing tasks, i.e., tasks whose output can be cached to serve requests from different consumers for a certain time. Caching such results, during their time validity, achieves the twofold benefit of reducing the service provisioning time and improving the edge resource utilization, by avoiding redundant computations and data exchange. The devised strategy is implemented as a network application of a Software-defined Networking Controller in charge of overseeing the edge domain. We formulate the optimal task placement through an integer linear programming problem, and we define an efficient heuristic algorithm that well approximates the solution achieved through a standard optimal solver. Achieved results show that the proposal successfully meets the targeted objectives in a wide variety of simulated scenarios, by outperforming benchmark solutions.

PushBox: Making Use of Every Bit of Time to Accelerate Completion of Data-Parallel Jobs

To minimize a job's completion time, we need to minimize the completion time of its final stage's last task. Scheduling of machine slots and networks largely dominates the variable part of each task's duration. Finding an optimal schedule is NP-hard even for offline and simplified scenarios. Previous work does lead to improved performance with various strategies. State-of-the-art task placement and network scheduling efforts are largely disjunctive. Without joint optimization, they are sub-optimal and myopic in many scenarios. Task placement usually treats the network as a black box. Thus, we use prioritized bandwidth allocation among tasks making the network both <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">predictable</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">efficient</i> to achieve joint scheduling. With this feature, joint scheduling can be transformed into a special <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">bin-packing problem</i> . Over this minimal yet power-enough abstraction, we propose PushBox to schedule data-parallel jobs in multi-tenant clusters. When designing the joint scheduling algorithm, we not only embrace the wisdom of prior art but also respect administrators’ fairness intent, which is so far largely ignored. We implement PushBox on Hadoop 3. PushBox performs persistently well on both a small testbed and a trace-driven simulator.

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.

Challenges for Implementing FAIR Digital Objects with High Performance Workflows

Challenges for Implementing FAIR Digital Objects with High Performance Workflows

A Model of the Test Technology of Teaching: Theoretical and Applied Aspects

The aim of this article is to reveal the results of studying the theoretical and applied aspects of the test technology of teaching: to structure the technological cycle, develop a model of the test technology of teaching, reveal the ways of optimisation of the educational process using the test technology. The conducted research shows that using test technologies of teaching significantly increases the qualitative indicators of mastering the material because of the motivational component, objectivity of evaluation, psychological comfort and elements of gamification. It has been determined that the efficiency of its implementation is affected by such aspects as compliance with the technological cycle, balancing the level of difficulty of test tasks by the difficulty index (Is), selection of the strategy for task placement, correctness of distractor creation, establishment of optimal quantitative and temporal characteristics of the test, the evaluation system, synchronous or asynchronous interaction between a teacher and students.

The impact of formulation of cost function in Task Mapping Problem on NoCs using bio-inspired based-metaheuristics

The Task Mapping Problem (TMP) of a Real-Time Application (RTA) onto a Many-/Multi-Processor System-on-a-Chip (MPSoC) with a Network-on-a-Chip (NoC) as on-chip communication architecture can be tackled as an optimization problem to improve desired design features in a static analysis setting. Examples of such features are the system’s compliance with its time requirements. A search-based bio-inspired metaheuristic such as the Genetic Algorithm (GA) can achieve task placement solutions in which all tasks are schedulable. However, the optimization processes are susceptible to the way the cost function is formulated. Additionally, solutions obtained with some of these methods do not consider some of the platform characteristics, such as the number of available virtual channels in the routers, and memory requirements in the processors, rendering their solutions unsuitable for real platforms. Therefore, we propose four objective functions to unify the end-to-end schedulability analysis that considers such platform characteristics. To demonstrate the performance of our objective functions, we conduct a series of experiments mapping the tasks of a benchmark application together with synthetic generated ones onto multiple Network-on-Chip platforms with realistic characteristics. We compare the objective functions during the optimization process using four different metaheuristics, showing that the multi-objective formulation helps the task mapper to achieve solutions that are more suitable for implemented NoC platforms. Although our study focuses on homogeneous NoC-Based MPSoCs, results have generated a better understanding of how the formulation of cost functions impacts the optimization processes for the specific TMP case, considering the Multi-Objective Optimization (MOO) context.