Optimal Power and Performance Management for Heterogeneous and Arbitrary Cloud Servers

The cost‐performance tradeoff is a fundamental issue in a data center for cloud computing, which is closely related to two key metrics that both cloud consumers and service providers care the most, that is, quality of service and cost of service. While there are different definitions of quality of service, the average response time is a common choice of performance metric. While there are various considerations in cost of service, the average power consumption is a common choice of cost metric. Hence, the cost‐performance tradeoff becomes the power‐performance tradeoff. In this article, we deal with the power‐performance tradeoff at the data center level. We study cost‐performance ratio optimization by using the techniques of workload management and server speed setting. In particular, we make the following tangible contributions. We solve three optimization problems, that is, (1) the workload management problem—to find a workload distribution, such that the cost‐performance ratio is minimized; (2) the server speed setting problem—to find a server speed setting, such that the cost‐performance ratio is minimized; (3) the workload management and server speed setting problem—to find a workload distribution and a server speed setting, such that the cost‐performance ratio is minimized. All the three optimization problems are analytically defined as multivariable optimization problems based on M/M/m queueing systems for multiple heterogeneous multiserver systems, together with two power consumption models, that is, the idle‐speed model and the constant‐speed model. Our approach makes it possible to quantitatively evaluate and optimize the cost‐performance ratio of a data center within a rigorously developed framework. Each multivariable optimization problem is transformed to a nonlinear system of equations. Due to the sophistication of these equations, they are solved algorithmically by a numerical procedure. Furthermore, we provide approximate, accurate, and analytical solutions to the first two problems. Performance data are demonstrated for each problem, and the accuracy of our approximate solutions are also discussed. To the best of the author's knowledge, this is the first paper which analytically and algorithmically minimizes the cost‐performance ratio of a data center with multiple heterogeneous multiserver systems using the techniques of workload management and server speed setting.

Real-time applications may be comprised of both soft and hard tasks, the former being related to streaming-based or other non-critical services. For these applications one is interested in minimizing the response time of soft tasks, for which quality of service is at stake, while not jeopardizing hard deadlines. In this paper, we propose a framework for multiprocessor real-time systems, that can schedule hard and soft tasks with unconstrained deadlines according to two EDF-based configurations. Our solution relies on two special servers, which are responsible for providing temporal isolation and slack reclaiming. These strategies are combined so that the average response time of soft tasks are improved. Results are evaluated via extensive simulation, which indicates that: the available processing capacity can be effectively utilized; average response time of aperiodic tasks is significantly reduced and hard deadlines are preserved while soft deadline miss ratio is kept low.

The detection of early-warning signals of performance deterioration can help technical support teams in taking swift remedial actions, thus ensuring rigor in production support operations of large scale software systems. Performance anomalies or deterioration, if left unattended, often result in system slowness and unavailability. In this paper, we presents a simple, intuitive and low-overhead technique for recognizing the early warning signs in near real time before they impact the system The technique is based on the inverse relationship which exists between throughput and average response time in a closed system. Because of this relationship, a significant increase in the average system response time causes an abrupt fall in system throughput. To identify such occurrences automatically, Individuals and Moving Range (XmR) control charts are used. We also provide a case study from a real-world production system, in which the technique has been successfully used. The use of this technique has reduced the occurrence of performance related incidents significantly in our daily operations. The technique is tool agnostic and can also be easily implemented in popular system monitoring tools by building custom extensions.

Cooling power constitutes a large portion of the total electrical power consumption in data centers. Approximately 25%∼40% of the electricity used within a production data center is consumed by the cooling system. Improving the cooling energy efficiency has attracted a great deal of research attention. Many strategies have been proposed for cutting the data center energy costs. One of the effective strategies for increasing the cooling efficiency is using dynamic thermal management. Another effective strategy is placing cooling devices (heat exchangers) closer to the source of heat. This is the basic design principle of many hybrid cooling systems and liquid cooling systems for data centers. Dynamic thermal management of data centers is a huge challenge, due to the fact that data centers are operated under complex dynamic conditions, even during normal operating conditions. In addition, hybrid cooling systems for data centers introduce additional localized cooling devices, such as in row cooling units and overhead coolers, which significantly increase the complexity of dynamic thermal management. Therefore, it is of paramount importance to characterize the dynamic responses of data centers under variations from different cooling units, such as cooling air flow rate variations. In this study, a detailed computational analysis of an in row cooler based hybrid cooled data center is conducted using a commercially available computational fluid dynamics (CFD) code. A representative CFD model for a raised floor data center with cold aisle-hot aisle arrangement fashion is developed. The hybrid cooling system is designed using perimeter CRAH units and localized in row cooling units. The CRAH unit supplies centralized cooling air to the under floor plenum, and the cooling air enters the cold aisle through perforated tiles. The in row cooling unit is located on the raised floor between the server racks. It supplies the cooling air directly to the cold aisle, and intakes hot air from the back of the racks (hot aisle). Therefore, two different cooling air sources are supplied to the cold aisle, but the ways they are delivered to the cold aisle are different. Several modeling cases are designed to study the transient effects of variations in the flow rates of the two cooling air sources. The server power and the cooling air flow variation combination scenarios are also modeled and studied. The detailed impacts of each modeling case on the rack inlet air temperature and cold aisle air flow distribution are studied. The results presented in this work provide an understanding of the effects of air flow variations on the thermal performance of data centers. The results and corresponding analysis is used for improving the running efficiency of this type of raised floor hybrid data centers using CRAH and IRC units.

The emergence of computation intensive automotive applications poses significant challenges on computation capacity of automotive electronic systems, thus vehicular edge computing (VEC) is introduced as a new computing paradigm into the internet of vehicle (IoV) to improve its data processing capability. However, as the computation capacity is limited in VEC servers, efficient task offloading algorithms need to be proposed. This paper first proposes a multi-task offloading model and gives the related task response time analysis method, and then, both a mixed-integer linear programming (MILP)-based algorithm and a simulated annealing-based heuristic algorithm are proposed to minimize the task response time. By comparing with a baseline algorithm, the MILP-based offloading algorithm can reduce the average task response time by 91.45%, and the heuristic offloading algorithm can reduce the average task response time by 70%.

In this letter, we analytically evaluate the average query response time of the Two-level Hierarchical Clustering based Hybrid-routing Protocol (THCHP) proposed recently for wireless sensor networks (WSNs). The accuracy of the average query response time analysis is verified via numerical simulations. Additionally, average query response time comparisons, between THCHP and a previously proposed Adaptive Periodic Threshold-sensitive Energy Efficient sensor Network (APTEEN) protocol, show that THCHP is better suited than APTEEN for delay sensitive WSN applications.

Today's data centers tend to have tens to hundreds of thousands of servers to provide massive and sophisticated services. Statistically, data center and data center networks (DCNs) remain highly underutilized which can be exploited for energy-saving. In this paper, we have studied energy-saving problem for network and server sides of the data center. The problem was formulated as a Mixed Integer Linear Program (MILP) that is solvable by an optimization software to jointly minimize the energy consumed by the servers and DCN. To overcome the optimization software high computational time, a heuristic algorithm to provide practical and efficient solution is introduced. The heuristic algorithm has two stages: first, it uses the virtual machines (VM) and the predicted servers resource utilization to provide VM consolidation algorithm and turn-off unused servers. The second stage uses an abstract performance-aware network flow consolidation that focused the traffic on subset of the network and turn-off the unused network devices. Simulation experiments using CloudsimSDN were conducted to validate the heuristic using real traces from Wikipedia in terms of energy consumption and average response time. The results show that the heuristic can save servers and network energy while maintaining performance.

In order to realize reliable and energy-aware server clusters, application processes have to be reliably and energy-efficiently performed in presence of server faults. In one way, an application process can be redundantly performed on multiple servers. In the improved redundant power consumption laxity-based (IRPCLB) algorithm which is proposed in our previous studies, since each application process is redundantly performed on more than one server, the reliability and availability of the cluster increase. However, the larger amount of electric power is consumed than non-redundant execution. In the IRPCLB algorithm, once a process terminates on a server, meaningless redundant processes on the other servers are forced to terminate in order to reduce the total power consumption. In this paper, we newly discuss the extended IRPCLB (EIRPCLB) algorithm to furthermore reduce the total power consumption of a server cluster. Here, in addition to forcing meaningless processes to terminate, each server is made reduce the consumption of computation resource to perform a redundant process by differentiating starting time of each redundant process. We evaluate the EIRPCLB algorithm in terms of total power consumption of a cluster and the average response time and number of servers to perform each process compared with the IRPCLB algorithm.

With the surge in data volume and dimensions, existing decision-making systems are difficult to capture nonlinear relationships and cope with the complexity and diversity of data, which leads to decision lag. To solve this problem, this article studies an intelligent decision-making system. The system integrates multi-layer feedforward neural networks and decision tree algorithms, and designs a distributed computing architecture based on the Apache Spark platform. Through the integration of the big data platform, the system integrates and processes multiple data sources; then, a three-layer feedforward neural network model is adopted, and a specific mechanism is used to prevent overfitting. After that, the decision tree algorithm is used to deeply explore the complex relationships in the data; Apache Spark is applied to accelerate the calculation process and efficiently query and analyze the stored data. The system also supports online learning strategies to continuously update the model to adapt to new data changes. The research results show that in the field of financial risk analysis, the average accuracy of the intelligent decision-making system in this article is 92.29 %, and the average response time is only 0.013 seconds; the average accuracy and response time of the traditional system are 82.54 % and 0.034 seconds respectively; the system in this article is significantly better than the traditional system. According to the study, an intelligent decision-making system not only increases decision-making's efficiency and accuracy but also offers solid base for more intelligent decisions in the future.

OEA-scheduler is a scheduler that generates a schedule of aperiodic tasks and periodic tasks by the principle of eliminating unnecessary context switches under the rate monotonic algorithm. This new scheduler will minimize the average response time of aperiodic tasks and also guarantee the hard deadlines of an arbitrary number of periodic tasks in hard real-time environments. In previous research, the OE-scheduler was developed to schedule periodic tasks only, and later improved to use object-oriented methods. Due to the object-oriented approach, the authors have been able to reuse the OE-scheduler algorithm and extend it to work with both periodic tasks and aperiodic tasks. An object oriented design of the scheduler will be presented, along with an implementation result

Uptime, operating cost and serviceability are the major concerns for data center cooling. This paper will describe an air and liquid cooling system that is optimized for data centers and does not involve major infrastructure changes. This system uses a minimum heat exchanger volume and coolant volume in order to simplify the installation at the server and data center level. The goal is to reduce chiller and server power, and removing heat from the CPU and the air in the server achieves that goal. Secondary goals include reducing the volume of liquid inside the server and making it easy to swap out racks, servers or components. The proposed system uses a normal base plate and fin heat sink, with a high performance liquid cooled heat exchanger built into it. This allows heat to be extracted from the CPU as well as the air that flows through the heat sink, removing up to 100% of the heat from the entire server with the liquid cooled heat sink which makes it a literal ‘heat sink’.

Background: Misoprostol is an analogue of prostaglandin (PGE1) and the first registered PGN used for the treatment of peptic ulcer disease. Misoprostol should not be used in pregnant women but it can be used in termination of intrauterine fetal death (IUFD). It has been used for the induction of abortion in the first and second trimester of pregnancy. Misoprostol has already been proved to be effective and efficient for second trimester since 1987. There is insufficient information available from clinical trials to support their use to induce labor in third trimester of pregnancy. Aim of the study: The aim of this study was to describe the effect of misoprostol in third trimester of pregnancy. Methods: This retrospective study was carried out in two large hospitals in Dhaka city, which is the capital of a South Asian country, Bangladesh. The record of the studied subject was taken from the Department of Obstetrics and Gynecology of Alpha Hospital and the maternity ward of Beta Hospital from June 2018 to August 2019. The actual names of the two hospitals are pseudonym for privacy. A detailed history of each patient was taken and physical and obstetric examination were done. In total 52 patients were investigated and data were analyzed by SPSS version 23.0. Results: In this study, the primi gravida response time was found as the highest. On an average 20.34 hours was needed for the primi gravida category. In single, 2 and multi-parity cases the average pain response times were found as 16.3125, 10.3375 and 14 hours respectively. In Primi, 2nd, 3rd and 5th gravida cases the average delivery times were found 13.0714, 12.5938, 9.0769 and 3 hours respectively. Against single, two and three doses of misoprostol, the average pain response times were found 7.6875, 11.5375 and 19.25 hours respectively. Conclusion: In this study, it had been found that, most of the patient of IUFD was ‘primigravid’ (30.8%) of total respondents having complication with medical and endocrinal disease. Pre- eclampsia is more common in prim gravid than multi. There is history of thyroid dysfunction. In some cases, now a day, GDM is the cause behind IUFD.

In Desktop Grid, managing faults is very crucial and challenging job. So a fault tolerant system is mandatory requirement in desktop grid for fault identification and their resolution. A fault tolerant system allows applications to continue execution despite having faults without termination. In this paper,an adaptive fault tolerant scheduling system is presented that combines dynamic replication with rescheduling. The system initially schedules jobs depending upon the completion time and fault rate of resources and then fault-tolerant strategies are applied depending upon availability of resources. To measure the performance,experiments has been carried out and it has been observed that proposed scheduling system outperforms by a factor of 4.8% w.r.t. Average task response time and 0.02% w.r.t. Average flowtime as compared to existing system. On the other hand, there is no significant improvement is observed when BoT completion time and average execution time are compared to existing system.

A large portion of the power consumption of data centers can be attributed to cooling. In dynamic thermal management mechanisms for data centers and servers, thermal setpoints are typically chosen statically and conservatively, which leaves significant room for improvement in the form of improved energy efficiency. In this paper, we propose two hierarchical thermal-aware power optimization techniques that are complementary to each other and achieve (i) lower overall system power with no performance penalty or (ii) higher performance within the same power budget. At the data center level, we trade off facility Heating, Ventilation and Air Conditioning (HVAC) power with server fan power by choosing between two thermal setpoints for the HVAC chiller based on the cooling zone utilization levels. This optimization can reduce total data center total power by as much as 12.4%-17%, with no performance penalty. At the server level, we trade off fan power and circuit leakage power by dynamically adjusting the server thermal setpoint, allowing the system to heat up when this saves more fan power than it costs in terms of leakage power. We evaluate this optimization on an IBM POWER 750 and find that it reduces total server power by up to 5.4% with no performance penalty for workloads that heavily exercise a server.

Although an Ultra Dense deployment is required for 5G services, it will be nearly impossible to achieve 60 service coverage with the dense deployments due to the even shorter transmission range. In light of 5G’s impressive technological advance, 5G and data centers are now capable of handling an increasing number of real-time and complicated computational tasks from Internet of Things (IoT) systems. This paper proposes an optimal mobile resource-sharing approach to confront this underlying limitation of 5G. In contrast to conventional algorithms, the designed optimal path planning and scheduling for mobile edge server (OPPSMES) is proposed that have the advantage of a synchronization among request being received and achieved lower delay and resource demand for as computing this allowed for the parallel processing of task and server in mobile condition. The OPPSMES includes two steps, i.e., path planning and optimal task scheduling, to improve efficiency. According to simulation outcomes, there is a significant increase in resource utilization and a decrease in average response time.