A Data-Driven Based Energy Consumption Modeling for Heterogeneous Servers in Data Centers

KEYWORD: Data center; DC power supply; PUE; energy efficiency ABSTRACT: Data center energy consumption is increasing with the rapid development of data cen- ter in recent years. This paper firstly elaborated data center energy saving technologies from three aspects: the power supply system, the air conditioning system and the IT epuipments, then carried on the energy efficiency comparative analysis of high voltage DC(HVDC) power supply technology and the traditional UPS system power supply technology. We come to the conclusion that the high vol- tage DC power supply technology obtained the better energy efficiency than the traditional UPS sys- tem power supply technology. A metric used to determine the energy efficiency of data center Before the discussion of the data center's energy efficiency, the metric used to measure the ener- gy efficiency of data center should be determined. Generally, EUI (Energy use intensity) was used to determine the enegy efficiency for common buildings. However, there will be some problems if EUI was used to determine the enegy efficiency of data center. Because of the wide variety and dif- ferent lay out of IT equipments, there will be a big difference in heat dissipation per unit area of IT equipment in different data centers. In this context, the EUI metric lost its significance, a new me- tric is needed to determine the energy efficiency of data center. The energy consumption of data centers is mainly distributed in IT equipment, air conditioning system, power supply and distribution system loss, etc. According to the statistical analysis of Ameri- can National Environmental Protection Bureau, the typical data center's energy consumption struc- ture is shown in Figure 1.

To determine whether a High-Performance Computing (HPC) data center is energy efficient, various aspects have to be taken into account: the data center's power distribution and cooling infrastructure, the HPC system itself, the influence of the system management software, and the HPC workloads; all can contribute to the overall energy efficiency of the data center. Currently, two well-established metrics are used to determine energy efficiency for HPC data centers and systems: Power Usage Effectiveness (PUE) and FLOPS per Watt (as defined by the Green500 in their ranking list). PUE evaluates the overhead for running a data center and FLOPS per Watt characterizes the energy efficiency of a system running the High-Performance Linpack (HPL) benchmark, i.e. floating point operations per second achieved with 1 watt of electrical power. Unfortunately, under closer examination even the combination of both metrics does not characterize the overall energy efficiency of a HPC data center. First, HPL does not constitute a representative workload for most of today's HPC applications and the rev 0.9 Green500 run rules for power measurements allows for excluding subsystems (e.g. networking, storage, cooling). Second, even a combination of PUE with FLOPS per Watt metric neglects that the total energy efficiency of a system can vary with the characteristics of the data center in which it is operated. This is due to different cooling technologies implemented in HPC systems and the difference in costs incurred by the data center removing the heat using these technologies. To address these issues, this paper introduces the metrics system PUE (sPUE) and Data center Workload Power Efficiency (DWPE). sPUE calculates the overhead for operating a given system in a certain data center. DWPE is then calculated by determining the energy efficiency of a specific workload and dividing it by the sPUE. DWPE can then be used to define the energy efficiency of running a given workload on a specific HPC system in a specific data center and is currently the only fully-integrated metric suitable for rating an HPC data center's energy efficiency. In addition, DWPE allows for predicting the energy efficiency of different HPC systems in existing HPC data centers, thus making it an ideal approach for guiding HPC system procurement. This paper concludes with a demonstration of the application of DWPE using a set of representative HPC workloads.

Data centers have become a critical computing infrastructure in the era of cloud computing. Temperature monitoring and forecasting are essential for preventing overheating-induced server shutdowns and improving a data center's energy efficiency. This paper presents a novel cyber-physical approach for temperature forecasting in data centers, which integrates Computational Fluid Dynamics (CFD) modeling, in situ wireless sensing, and real-time data-driven prediction. To ensure the forecasting fidelity, we leverage the realistic physical thermodynamic models of CFD to generate transient temperature distribution and calibrate it using sensor feedback. Both simulated temperature distribution and sensor measurements are then used to train a real-time prediction algorithm. As a result, our approach significantly reduces the computational complexity of online temperature modeling and prediction, which enables a portable, noninvasive thermal monitoring solution that does not rely on the infrastructure of monitored data center. We extensively evaluated our system on a rack of 15 servers and a test bed of five racks and 229 servers in a production data center. Our results show that our system can predict the temperature evolution of servers with highly dynamic workloads at an average error of 0.52C, within a duration up to 10 minutes.

Recently the huge amount of energy consumption has become a barrier to the widespread deployment of data centers serving various Internet of Things applications. The reasonable allocation of compute-intensive workloads to physical servers is an efficient way to improve the data center's energy efficiency. Though existing works has proposed some algorithms to manage workloads or virtual machines for energy saving, most of them did not comprehensively consider the high dynamics of server states, and lacked in high scalability in their implementation. In this paper, the Actor Critic based Compute-Intensive Workload Allocation Scheme (AC-CIWAS) is proposed, which can both guarantee the Quality of Service (QoS) of workloads and reduce the computational energy consumption of physical servers. To achieve rational workload allocation, AC-CIWAS captures the dynamic feature of server states continuously, and takes the impact of different workloads on energy consumption into consideration. AC-CIWAS employs the Deep Reinforcement Learning (DRL) based Actor Critic (AC) algorithm to evaluate the expected cumulative return over time, while the cumulative return guides to allocate workloads with high energy efficiency. Simulation results have demonstrated that compared to existing baseline allocation methods, the proposed AC-CIWAS can achieve an approximately 20 percent decrease in server power consumption with QoS guarantee.

An Increase in Data Center power requirements has placed significant pressure on traditional Data Center cooling management Systems. The temperature in the Data Center is controlled using Air handling units (AHUs) and plays a critical role in a Data Center to maintain the required temperature to ensure the best possible performance. As the targeted Data Center is quite Old and using backdated technologies and does not have sensor-based technologies implemented. One of the issues faced by the target Data Center was that AHU fan speed was set to the static setting which impacts the Supplied Temperature in Data Center and results in excessive hot & cold temperature inside a Data Center. The proposed model resolves the problem faced by the targeted Data Center to operate the AHU fan speed to maintain the required DC Temperature on the predicted range by using machine learning techniques. This model not only solves the problem of maintaining the necessary temperature in the Data Center, but it can also regulate the fan speed within the expected range, contributing to the Data Center's energy efficiency. Supervised machine learning with linear regression and logistical regression approaches are utilized to investigate which methodology produces the best prediction results for adjusting the AHU unit fan speed for better control of the supplied Data Center temperature. In the targeted Data Center, it has no scope to expand more rack space or host IT load. It is desired that the predicted or recommended range for controlling AHU fan speed be determined so that the needed temperature can be sustained with the suggested setting without requiring extensive manual task. Henceforth as the data generated by the Data Center is historical, supervised regression machine learning models using Linear and Logistic Regression techniques are used. Both regression models are compared to see which regression methodology predicts the best variable fan speed range for maintaining the data center's required temperature.

Data centers have a huge impact on the world we live in. Today data centers account for 3% of the global electricity supply and consume more power than the total of some country consumption. Data centers also contribute 2% of the total global greenhouse gas emissions. Data center's energy efficiency is the eternal hot research and engineering hot topic. In this paper, an advanced distributed power backup battery power system (BBS) design is introduced, also engineering comparison between lead-acid battery (VRLA), lithium iron phosphate battery (LFP) and ternary lithium-ion battery (NCM) is detailed listed with key parameters. Moreover, this paper has proved the best engineering practice of using the cylindrical lithium iron phosphate (LiFePO4) battery for BBS design optimization considering the reliability, cost saving, and energy efficiency of data center deployment in a large scale. Engineering work summarized in this paper includes a serial of related tests of the electrical performance, life cycle and reliability items, and the corresponding test results reaching design targets of power supply capacity, energy efficiency improvement for data center which eventually contributed a total cost ownership (TCO) saving. The theoretical basis, proto product, engineering test result and deployment analysis of LiFePO4 battery backup power system is data-proved reference for ecosystem in their new data center construction.

Data centre's energy efficiency optimization and greening—Case study methodology and R&D needs

Benchmarking the data centre's energy efficiency is a key step towards optimising and reducing the power consumption and the related energy costs. Power usage effectiveness PUE was coined by the members of American Green Grid as an index to measure the energy efficiency of data centres. Traditional approaches of calculating the PUE in data centres only consider the impact of the weighted time, but ignore the impact of the corresponding workload. These methods cannot accurately reflect the long-term energy efficiency of data centres. Since the power consumed by IT equipments and the temperature generated by the equipments normally grow with the increase of the workload intensity, an effective method employed to calculate the PUE of data centres has to consider the impacts of workload. In contrast to the traditional methods, this paper proposes an approach to compute the PUE by leveraging the workload fluctuation. Theoretical analysis and experimental evaluation demonstrate that the proposed approach can achieve a more accurate PUE than that of the traditional methods.

To cut back energy consumption of virtual-machine-powered data centers, we build an optimization model for virtual machines running in DVFS-enabled cloud data centers. With the model in place, cloud computing systems are equipped to keep track of dynamic power and static power of processors in virtual machines. Unlike existing dynamic voltage and frequency scaling schemes, our solution orchestrates frequency requirements rather than task execution times. The model makes it possible to obtain an optimal frequency ratio, which minimizes energy consumption of virtual machines. As a result, a data center's energy efficiency is boosted by controlling CPU frequency to meet the optimal frequency ratio. We demonstrate a way of manipulating frequency ratios to pushing up energy efficiency without violating virtual machines' frequency requirements. The experimental results unveil that our modeling approach offers a practical way of conserving the energy consumption of virtual machines running in data centers.

Energy-efficient virtual machine placement in data centres via an accelerated Genetic Algorithm with improved fitness computation

Enhancing the efficiency and the reliability of the data center are the technical challenges for maintaining the quality of services for the end-users in the data center operation. The energy consumption models of the data center components are pivotal for ensuring the optimal design of the internal facilities and limiting the energy consumption of the data center. The reliability modeling of the data center is also important since the end-user’s satisfaction depends on the availability of the data center services. In this review, the state-of-the-art and the research gaps of data center energy consumption and reliability modeling are identified, which could be beneficial for future research on data center design, planning, and operation. The energy consumption models of the data center components in major load sections i.e., information technology (IT), internal power conditioning system (IPCS), and cooling load section are systematically reviewed and classified, which reveals the advantages and disadvantages of the models for different applications. Based on this analysis and related findings it is concluded that the availability of the model parameters and variables are more important than the accuracy, and the energy consumption models are often necessary for data center reliability studies. Additionally, the lack of research on the IPCS consumption modeling is identified, while the IPCS power losses could cause reliability issues and should be considered with importance for designing the data center. The absence of a review on data center reliability analysis is identified that leads this paper to review the data center reliability assessment aspects, which is needed for ensuring the adaptation of new technologies and equipment in the data center. The state-of-the-art of the reliability indices, reliability models, and methodologies are systematically reviewed in this paper for the first time, where the methodologies are divided into two groups i.e., analytical and simulation-based approaches. There is a lack of research on the data center cooling section reliability analysis and the data center components’ failure data, which are identified as research gaps. In addition, the dependency of different load sections for reliability analysis of the data center is also included that shows the service reliability of the data center is impacted by the IPCS and the cooling section.

Our research aims to improve energy efficiency in data centres by combining cloud computing infrastructure with machine learning techniques. We propose that dynamic resource assignment, combined with intelligent optimisation of cooling systems, can reduce power waste and operational costs. Real-time sensor data from various data centre components such as servers, cooling systems, and power distribution units is collected and fed into machine learning models for analysis. In this way, we can create power arrangements that are tailored to the resources and needs of various applications. The experimental results show that energy consumption has been reduced by an average of 30% compared to traditional methods. Furthermore, our machine learning models are quite accurate in predicting cooling system performance. For example, an Artificial Neural Network (ANN) has an accuracy rate of 98.78%. This result demonstrates the efficacy of our approach in promoting energy efficiency and operational performance in data centres: it not only provides a scalable, cost-effective solution to industry energy efficiency challenges, but it also improves day-to-day data centre operations by reducing electrical consumption. Our approach, which is based on dynamic allocation of computational resources and real-time data analysis for optimising cooling systems, not only saves energy but also improves operational efficiency. With energy in mind, we must all work towards a more sustainable and green approach to data centre management. Future research can look into other potential optimisations, as well as issues with scalability and application in real-world data centre environments.

Climate change is recognised as one of the key challenges humankind is facing. The Information and Communication Technology (ICT) sector including data centres generates up to 2% of the global CO2 emissions, a number on par to the aviation sector contribution, and data centres are estimated to have the fastest growing carbon footprint from across the whole ICT sector, mainly due to technological advances such as the cloud computing and the rapid growth of the use of Internet services. There are no recent estimations of the total energy consumption of the European data centre and of their energy efficiency. The aim of this paper is to evaluate, analyse and present the current trends in energy consumption and efficiency in data centres in the European Union using the data submitted by companies participating in the European Code of Conduct for Data Centre Energy Efficiency programme, a voluntary initiative created in 2008 in response to the increasing energy consumption in data centres and the need to reduce the related environmental, economic and energy supply security impacts. The analysis shows that the average Power Usage Effectiveness (PUE) of the facilities participating in the programme is declining year after year. This confirms that voluntary approaches could be effective in addressing climate and energy issue.

Issues in Energy Efficiency of Data Centers (DC) are important, due to the cumulative effects of the increase in the DCs number and in the energy consumption per center. Developing new design recommendations to improve a cooling system efficiency, commonly quantified by the PUE metric (Power Usage Effectiveness) is one objective of the Green IT organizations. For existing DCs, without considering the optimization of the IT workload, a possible way to improve the DC's energy efficiency is to adjust the cooling setpoints. In this paper, a methodology based on predictive models is used to optimize the PUE by improving the cooling setting. The modeling approach consists in exploiting the temperatures and energy measurements at various operating conditions to predict the PUE behavior using data-driven models commonly called black box models. The optimization procedures are based on the simulation of these models in order to estimate the best working conditions.

this paper puts forward the monitoring and measurement methods on energy consumption of virtual machine in the cloud data center, establishes energy consumption model of virtual machine system and virtual machine migration. The usual migration method of virtual machine uses heuristic algorithm to allocate virtual machine, its solution result is easily to be got into locally optimal solution, and this paper gives migration algorithm of virtual machine based on genetic algorithm on the basis of making research on genetic algorithm, it makes improvement on the target function in genetic algorithm, making application number of target node and migration time minimum on condition of meeting protocol of service class, so that it can realize energy conservation in data center. Introduction With the popularization and quick development [1, 2,3] of computer technology, the task request reaches in cloud computing platform becomes to be diversified. In order to meet task demand of different kinds, the calculation node in cloud data center constitutes hardware platform of cloud calculation has to keep open state for long time and wait for arrival of task, which will cause low application and high waste of cloud data center on energy consumption. Energy consumption model can be said to be one of the most important parts in cloud data center, when cloud data center is continually operating, we should make deep comprehension on users and administrators of cloud calculation, know their application way, so that make corresponding solution measures, so that it can reach target of optimization and energy conservation. At present, many servers of cloud data have the self-detection ability, they can measure some data components, but this single physical detection ability is obviously accords with future development idea of cloud data center, only quicker and effective energy conservation measures can make energy consumption modeling have cleanness . This paper tries to establish reasonable and reliable energy consumption model for cloud data center from layer of cloud infrastructure, and it compares with effect of different sampling ways and mathematical techniques on energy consumption model. On algorithm and test, it uses research outcome of energy consumption model put forward by this paper to demonstrate effectiveness of energy consumption model state by this paper, and this energy consumption model can be also applied to other research work on it. System structure on energy management of cloud computing It uses virtualized technology as core and fully considers resources application and load characteristics under cloud calculation environment, it makes effective monitoring management and optimization on energy consumption of cloud calculation, reduce total energy consumption of virtual cluster. It also provides calculation resources server on IT service, it can be indicated as DC = {H1,H2,...,HN}. It indicates calculation node, which is the physical server provides calculation ability, data center in distributed calculation, the node is divided into calculation node and storage node, data and file storage are in the storage node of NFS. The physical master is isostructural, it has the same calculation resources capacity(CPU, memory, disc), as the parasitifer, physical master International Conference on Applied Science and Engineering Innovation (ASEI 2015) © 2015. The authors Published by Atlantis Press 594 opertes isomeric virtual machine, it is provided by Vmware or XEN. Virtual machine is isomeric, bit it is not random isomerism, it has certain virtual machine category defined in advance VC(vm class) = {VCi,-,VCs], virtual machine of every category is isostructural, its resources occupancy is (resource)={R1,...,RS}. In order to record and describe state of current cluster, it adopts matrix to describe state in virtual cluster, the formula 1 is indicated as follows: