Dynamic Resource Provisioning and Scheduling with Deadline Constraint in Elastic Cloud

Purpose of ReviewTheoretical frameworks for adaptive natural resource management are quite common, whereas documented examples showing successful implementation of adaptive management and learning through multiple time intervals have remained uncommon. Measures of quality of adaptive natural resource management processes are needed to examine potential factors driving the successful implementation. To address this gap, we developed a multimetric index composed of 22 metrics to assess quality of case studies using quantitative decision support (QDS) to inform adaptive forest management (AFM). Metrics represented three main tasks, including conceptual setup, modeling, and application. We further distinguished these into subtasks: definition of objectives and management options (setup); specifying uncertainty, prediction, and optimization (modeling); and stakeholder involvement along with practice and learning (application). We used a multimetric index to examine temporal and geographic variation in quality of reviewed case studies using QDS to inform AFM. We then conducted a structured literature review of 179 articles, wherein 34 case studies met a priori criteria.Recent FindingsWhen applying the multimetric index to these case studies, we found that over the past decade the index has been intermediate and annual average scores declined by 33% from 4.5 to 3.0 of 10 (where 10 is the highest possible quality score). Aligning with reviews of adaptive natural resource management, reported on-ground application of QDS to inform AFM was rare (n = 2). We also confirmed the expectation that there has been a substantial lack of stakeholder engagement during QDS development tasks.SummaryOur multimetric index provides a novel tool to examine gaps in the use of QDS for adaptive management in diverse domains including but not limited to forests.

Cloud computing has become inevitable in own way with its offering multi-services, deployment, resource provisioning, data management, etc. Efficient resource consumption proves the dignity of a reliable cloud service provider in the market. However, the cloud environment often witnesses the uncertainties with respect to the resource provisioning. The most significant uncertainty encountered is the fault that paves way for a failure in the total process. Existing methods fail to incorporate such uncertainty into the picture leading to performance degradation of the cloud service requested by cloud user. The proposed scheme, robust fault detection and recuperation (RFDR), is a reactive fault-tolerant strategy that monitors the system continuously and recovers it from failures caused by VMs, host, and PEs. The proposed mechanism is run through baseline scheduling algorithm such as first come first service (FCFS), shortest job first (SJF), priority scheduling (PS), min-min, and max–min scheduling for the performance evaluation. The experiments indicate that the proposed method efficiently handles the fault detection and recover with 99% of resource availability and reliability.

Cloud computing is an IT paradigm, a model for enabling access to shared huge configurable resources, which is provisioned with minimal management effort, often over the web. Users access the cloud applications through a browser or a mobile app while the software and data are stored on servers at a remote location. The benefits of using the cloud are storage, security, cost, accessibility, flexibility, maintenance, backup and recovery etc. It also provides easy access to data at any time. Cloud application providers strive to give the better service and performance as if the software programs were installed locally on end-users system. One of major issue in cloud computing is resource allocation. Resource allocation is difficult to solve, especially in case of complex task and performance computing. To satisfy the performance, mapping the tasks to processor is done, which helps in resource utilization in the process efficiently. The existing Shortest Job First (SJF) algorithm uses the CPU processing time and memory size to perform the shortest job scheduling first. The starvation in SJF exists only if processes with low burst time appears in queue before the processes with the high burst time is executed. Since this algorithm always chooses the process with low burst time, the process with the high burst time will never get the share of CPU. So, in order to avoid starvation, the proposed system combines the SJF and the Multilevel Queue (MQ) Scheduling based on credit.

Modern CPUs suffer from performance and power consumption variability due to the manufacturing process. As a result, systems that do not consider such variability caused by manufacturing issues lead to performance degradations and wasted power. In order to avoid such negative impact, users and system administrators must actively counteract any manufacturing variability. In this work we show that parallel systems benefit from taking into account the consequences of manufacturing variability when making scheduling decisions at the job scheduler level. We also show that it is possible to predict the impact of this variability on specific applications by using variability-aware power prediction models. Based on these power models, we propose two job scheduling policies that consider the effects of manufacturing variability for each application and that ensure that power consumption stays under a system-wide power budget. We evaluate our policies under different power budgets and traffic scenarios, consisting of both single- and multi-node parallel applications, utilizing up to 4096 cores in total. We demonstrate that they decrease job turnaround time, compared to contemporary scheduling policies used on production clusters, up to 31% while saving up to 5.5% energy.

With the advent of High Performance Computing (HPC) in the large-scale parallel computational environment, job scheduling and resource allocation techniques are required to deliver the Quality of Service (QoS) and resource management. Therefore, job scheduling on a large-scale parallel system has been studied to: (a) minimize the queue time and response time, and (b) maximize the overall system utilization. We compare and analyze thirteen job scheduling policies to analyze their behavior. The set of job scheduling policies include: (a) priority-based policies, (b) first fit, (c) backfilling techniques, and (d) window-based policies. All of the policies are extensively simulated and compared. A real data center workload comprised of 22385 jobs is used for simulation. We analyze the: (a) queue time, (b) response time, and (c) slowdown ratio to evaluate the policies. Moreover, we present a comprehensive workload characterization that can be used as a tool for optimizing system's performance and for scheduler design. We investigate four categories of the workload characteristics including: (a) Narrow, (b) Wide, (c) Short, and (d) Long 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.

Adaptive resource management (ARM) is an approach to managing that allows decision makers to learn about a system and subsequently change management actions based on new information about system processes (i.e., adapt) to better meet fundamental objectives. This is not to be confused with changing management actions when the state of the system changes. For example, changing a harvest regulation when populations decline is not ARM. This dynamic decision making may be fortuitously optimal, but if the effect of harvest is uncertain then changing regulations may be suboptimal—for example, weather may have caused the decline. Adaptive resource management can be implemented along a spectrum of passive to active to reduce system uncertainty. Active ARM is when explicit hypotheses are posited then implemented to test them, and monitoring occurs to elucidate whether the effect of the management action achieved a given result. Passive ARM uses current management practices, natural variation in the system, and monitoring to reduce uncertainty. Even though northern bobwhite (Colinus virginianus) have been studied for a century, uncertainty regarding optimal management strategies still exists. The Albany Quail Project (AQP) has used both modes of ARM to learn about northern bobwhite populations to better meet the hunting objectives of stakeholders. Between 1992 and 2019 the AQP radio-tagged 5,182 unique individuals and banded an additional 5,008 birds on the primary study area near Albany, Georgia, USA. Additionally, 1,724 nests have been monitored and population surveys conducted in the spring and autumn. Active ARM occurred with tests of supplemental feeding, hardwood cleanup, and predator control whereas passive ARM was used to learn about prescribed fire regimes, brood habitat preferences, small mammal cycles, and raptor migrations. We built an integrated population model (IPM) that combined known-fate survival data, nesting records, dead recoveries from harvest, and population surveys to model the system. The preliminary results of the IPM demonstrated that populations have increased during the project. The 5 most abundant autumn populations have occurred in the most recent 10 years of the project when bobwhite populations throughout their range have continued to decline. Unlike typical ARM applications, the AQP has not had an explicit model to predict consequences of future management actions; nonetheless, the essence of ARM has been followed. Since its inception uncertainty has been reduced, management actions have been changed, and bobwhite populations have responded. Improvements to AQP and similar projects could include explicit hypotheses and predictive models about the system

Cloud Computing provides services, on-demand access, infrastructure, storage of data and application. It possesses the reliability, availability and the scalability. One of the issues in cloud computing is Energy Saving. In this paper, the proposed work is Energy Saving Task Scheduling (ESTS) methodology. The aim of this methodology is to show the performance comparison of all the task scheduling types. Task scheduling or Job scheduling is referred to as policies that control the work order to be performed by a computer system. Types of Task Scheduling are Shortest Job First (SJF), First Come First Serve (FCFS), Round Robin (RR) and Priority Scheduling. In each type of schedule, the processes used by the parameters were calculated. Finally, the performance comparison is made in scheduling algorithms and shows better results. This method is implemented in net beans toolkit.

Resource management strategy in adaptive cloud provisions the needed resources dynamically to the end-users. To improve the runtime performance of adaptive cloud for service-based applications, two aspects of technical issues are required to be addressed. The first one is the balancing of a large amount of data on existing resources and the second is resource provisioning which can adjust the number of resources optimally to adapt the time-varying workload. As the growth of data is increasing tremendously, efficient resource management is the need in cloud computing. We build a cloud framework to process data in automation with adaptive resource and workload management strategy. Numbers of approaches are reviewed and applied for workload prediction. We developed the model Auto-Regressive Integrated Moving Average-workload Prediction for Efficient Resource Provisioning (ARIMA-PERP) and evaluated the results that can satisfy the on-demand need of end-users for efficient resource utilization. To serve the maximum number of user requests, performance metrics of the proposed approach are evaluated. It is observed that our evaluated results achieved an accurate prediction by 91.11%, which meets the efficient resource utilization for the demanded workload. As compared with the exiting approach, we achieved better performance by 0.11% for accurate prediction. The proposed architecture is intended to provide the resources dynamically and efficiently satisfying the demands of the user. To achieve this objective of efficient resource provisioning, algorithms are developed for workload prediction which helps in deciding optimum resource provisioning. Our system uses proactive approach resource management and deployment of the adaptive cloud system. In traditional systems, resources are managed based on demand, availability and the strategy of scheduling which results in delayed response time at large. We configured the ARIMA model to predict the future workload for provisioning the resources dynamically and remove the problem of over-provisioning and under-provisioning in a cloud environment.

Load balancing in cloud computing assigning the all load over each node of a mutual or shared system to make resource usage effective and to perform the response time to the job. This has used Round Robin (RR) and shortest job first (SJF) algorithm. Scheduling is the main part of cloud computing and jobs or workflow scheduling under the platform as a service. The work under infrastructure as a service work to virtual machine scheduling which machine is decided by scheduler should go on which job or virtual machine. Simply RR is a time quantum-based preemptive algorithm, and SJF is a burst time-based non-preemptive and preemptive algorithm. But RR work in two-phase like expected finish time adapted for each process. Each process with generally minimum completion times scheduled. This algorithm result as no longer waiting for the smallest task.

The Shortest Job First (SJF) algorithm reduces the average waiting time of processes in a CPU environment; however, it may lead to the starvation of long processes. The Round Robin (RR) algorithm solves starvation by equally distributing processing time. This paper explores user-friendly improvements by allowing users to specify which processes should be finished first. The SJF, RR, and the user-priority SJFRR were compared using four test cases in the following criteria: context switching, average waiting time, and average turnaround time. The user-priority SJFRR was found to have reduced average waiting time and average turnaround time in all test cases except for the first. Further research is also recommended in order to find out whether the algorithm's performance is attributed to the randomized assignment of priority and to find a proper basis for user designation of priority tasks.

Recent Infrastructure-as-a-Service offers, such as Amazon's EC2 cloud, provide virtualized on-demand computing resources on a pay-per-use model. From the user point of view, the cloud provides an inexhaustible supply of resources, which can be dynamically claimed and released. This drastically changes the problem of resource provisioning and job scheduling. This article presents how billing models can be exploited by provisioning strategies to find a trade-off between fast/expensive computations and slow/cheap ones for indepedent sequential jobs. We study a dozen strategies based on classic heuristics for online scheduling and bin-packing problems, with the double objective of minimizing the wait time (and hence the completion time) of jobs and the monetary cost of the rented resources. We simulate these strategies on real grid workloads in two cases. First, we use the workloads as a whole, which is representative of a large community of users sharing some common resources. Second, we use the workloads extracted for each individual user. These lighter workloads correspond to users submitting work independently from others and paying for their own resources. Our findings show that on large workloads, a little budget increase allows to achieve optimal wait time, while trade-off heuristics may be largely beneficial for individual users with lighter workloads.

Load sharing is a technique to improve the performance of distributed systems by distributing the system workload from heavily loaded nodes, where service is poor, to lightly loaded nodes in the system. Previous studies have considered two adaptive load sharing policies: sender-initiated and receiver-initiated. In the sender-initiated policy, a heavily loaded node attempts to transfer work to a lightly loaded node and in the receiver-initiated policy a lightly loaded node attempts to get work from a heavily loaded node. Almost all the previous studies assumed the first-come/first-served node scheduling policy; furthermore, analysis and simulations in these studies have been done under the assumption that the job service times are exponentially distributed and the job arrivals form a Poisson process (i.e., job inter-arrival times are exponentially distributed). The goal of this paper is to fill the void in the existing literature. We study the impact of these assumptions on the performance of the sender-initiated and receiver initiated policies. We consider three node scheduling policies-first-come/first-served (FCFS), shortest job first (SJF), and round robin (RR) policies. Furthermore, we also look at the impact of variance in the inter-arrival times and in the job service times. Our results show that: (i) When non-preemptive node scheduling policies (FCFS and SJF) are used, the receiver-initiated policy is (substantially) more sensitive to variance in inter-arrival times than the sender-initiated policies and the sender-initiated policies are relatively more sensitive to the variance in job service times; (ii) When the preemptive node scheduling policy (RR) is used, the sender-initiated policy provides a better performance than the receiver-initiated policy.

Real-time and embedded systems have traditionally been designed for closed environments where operating conditions, input workloads, and resource availability are known a priori and are subject to little or no change at runtime. There is an increasing demand, however, for autonomous capabilities in open distributed real-time and embedded (DRE) systems that execute in environments where input workload and resource availability cannot be accurately characterized a priori. These systems can benefit from autonomic computing capabilities, such as self-(re)configuration and self-optimization, that enable autonomous adaptation under varying-even unpredictable-operational conditions. A challenging problem faced by researchers and developers in enabling autonomic computing capabilities to open DRE systems involves devising adaptive planning and resource management strategies that can meet mission objectives and end-to-end quality of service (QoS) requirements of applications. To address this challenge, this paper presents the integrated planning, allocation, and control (IPAC) framework, which provides decision-theoretic planning, dynamic resource allocation, and runtime system control to provide coordinated system adaptation and enable the autonomous operation of open DRE systems. This paper presents two contributions to research on autonomic computing for open DRE systems. First, we describe the design of IPAC and show how IPAC resolves the challenges associated with the autonomous operation of a representative open DRE system case study. Second, we empirically evaluate the planning and adaptive resource management capabilities of IPAC in the context of our case study. Our experimental results demonstrate that IPAC enables the autonomous operation of open DRE systems by performing adaptive planning and management of system resources.

Job scheduling policies for HPC centers have been extensively studied in the last few years, especially backfilling based policies. Almost all of these studies have been done using simulation tools. All the existent simulators use the runtime (either estimated or real) provided in the workload as a basis of their simulations. In our previous work we analyzed the impact on system performance of considering the resource sharing (memory bandwidth) of running jobs including a new resource model in the Alvio simulator. Based on this studies we proposed the LessConsume and LessConsume Threshold resource selection policies. Both are oriented to reduce the saturation of the shared resources thus increasing the performance of the system. The results showed how both resource allocation policies shown how the performance of the system can be improved by considering where the jobs are finally allocated. Using the LessConsume Threshold Resource Selection Policy, we propose a new backfilling strategy : the Resource Usage Aware Backfilling job scheduling policy. This is a backfilling based scheduling policy where the algorithms which decide which job has to be executed and how jobs have to be backfilled are based on a different Threshold configurations. This backfilling variant that considers how the shared resources are used by the scheduled jobs. Rather than backfilling the first job that can moved to the run queue based on the job arrival time or job size, it looks ahead to the next queued jobs, and tries to allocate jobs that would experience lower penalized runtime caused by the resource sharing saturation. In the paper we demostrate how the exchange of scheduling information between the local resource manager and the scheduler can improve substantially the performance of the system when the resource sharing is considered. We show how it can achieve a close response time performance that the shorest job first Backfilling with First Fit (oriented to improve the start time for the allocated jobs) providing a qualitative improvement in the number of killed jobs and in the percentage of penalized runtime.

Elasticity is a key feature of current cloud computing platforms. Dependent on their demand tenants can dynamically scale up and down their applications. To increase their revenue, cloud providers are used to over-provision their clusters, but they still have to reserve capacity to avoid that services get unresponsive and cause SLO violation during bursts. In this paper, we propose CLOUDFARM, a PaaS architecture with an adaptive SLO-based resource management mechanism. It introduces new flexible SLAs backed with a respective development model and management interface for end-user services. According to their SLAs and the price tenants pay, services can be selectively downgraded to overcome short-term peaks, e.g. While preparing for scale-out. Providers can deploy services optimistically and thus maximize their data center utilization and revenue.