Search

With the development of the information and communication technology (ICT) industry, the energy consumption of data centers is continuously increasing. High-temperature polymer electrolyte fuel cells (HT-PEFC) have a high operating temperature of 120 °C or higher; thus, the heat generated from fuel cells stack is effectively used as a heat source for a absorption refrigerator (AR) to generate cooling. The combined cooling and power (CCP) system is proposed to satisfy the energy demand of data centers that require both power and cooling. The CCP system comprises an HT-PEFC and a double-effect AR that recovers the wasted heat of the fuel cell stack. As a result of analyzing the electrical and cooling efficiency of the CCP system according to the fuel cell operating conditions, the overall efficiency increased to 95%, which was significantly higher than the existing system. Based on the simulation of the developed model, coefficient of performance (COP) and cooling capacity depending on the temperature changes of chilled water and cooling water were calculated within the stack temperature range of 150–180 °C, and the COP changed from 1.1 to 1.58 depending on the conditions. The developed CCP system model can be used to plan strategies for flexible fuel cell operation according to the power and cooling demands required in the data center. • The combined cooling and power (CCP) system to meet energy demand of data centers. • The CCP system comprises an HT-PEFC and a double-effect absorption refrigerator. • The electric and cooling efficiency of the proposed CCP system were analyzed. [ABSTRACT FROM AUTHOR]

Cloud services have become a demand in big cloud data centers for allocating energy-efficient resources and resource utilization. Most current solutions consider a small number of resource categories, which causes an imbalanced load or the needless activation of real servers. Thus, the virtual machine placement-based bin padding algorithm was proposed and analyzed with the equilibrium state of multi-resource utilization. Additionally, the impact of imbalanced resources causes weak virtual machine task allocation that induces poor VMs' scheduling activities. Therefore, this work incorporates efficient conjectured scheduling to prioritize the task based on computational time with trust parameters calculation. Moreover, to improve the performance, the new abatement strategy was used with efficient resource management, which generates a physical host to minimize energy burning and increase the quality of service to the data center by reducing overload. Implementing the proposed algorithm efficiently utilizes the memory of up to 290 MB for processing 90 MB files with a low computational time of 23 s Further algorithms tested on ten benchmark functions show that they outperform similar algorithms. Thus, the proposed model applies to individual servers in the cloud. [ABSTRACT FROM AUTHOR]

Data centers store the necessary data for the operation of services and applications used by users. They usually consist of many physical servers connected to each other, with virtual machines and other systems, applications, databases installed. In order to increase the reliability of data access and to protect the system from damage or destruction, data from servers are replicated to other centers, servers and instances. The article examines the average replication time for whole virtual machines of various sizes and differential replication of virtual machines for exemplary data center. The research was conducted in order to investigate the influence of bandwidth constraints between data centers on replication times. The article presents sample results from a real system. [ABSTRACT FROM AUTHOR]

Heat propagation and servers' temperature increase inside data centers racks is a vital issue. So, selecting the proper cooling architecture is an important step during the design of data centers. Traditional cooling systems, called perimeter cooling, consume huge amounts of electricity. A more recent effective cooling architecture of data centers, called in-row cooling where the cooling units are inserted between racks, is commercially suggested to reduce cooling power consumption. The current study numerically investigates the performance of in-row cooling architecture compared with the traditional cooling architecture of data centers of different power densities. Temperature distribution and performance parameters indices such as Supply/Return Heat Indices (SHI/RHI), Return Temperature Index (RTI), Index of Mixing (IOM), Energy Utilization Coefficient (ηr), and Beta Index (β) are used to conduct this comparative study. The study was performed at different rack's power densities to determine the overall better cooling architecture for the different power densities. The results show that (i) in-row cooling architecture has better thermal performance as observed in temperature contours (ii) SHI, RHI, RTI, ηr, β and IOM have better values in case of in-row cooling, especially at high power densities, which indicates less hot air recirculation and cold air bypass and completely benefit from cooling capacity of CRACs, (iii) perimeter cooling is suitable for low power densities while in-row cooling can be used for high power densities, and (iv) maximum velocity values are obtained for 10 kW rack's power which dramatically affects the thermal performance of perimeter cooling architecture as indicated by the dramatic decrease of the energy utilization coefficient to 1.4 while it increased for in-row cooling to 1.9. [ABSTRACT FROM AUTHOR]

The widespread deployment of data centers has contributed to a dramatic increase in energy consumption and carbon emissions. To solve this problem, this study proposes a distributed multi-energy system (DMES) driven by solar, wind, geothermal and natural gas and its life-cycle-based multi-objective optimization model considered energy, economy, and environment. Moreover, a novel operation strategy based on load characteristics is presented for energy flow allocation and a multi-objective grasshopper optimization algorithm is improved for model solving. Case studies are conducted on a practical data center in Qinghai, China. Based on the DMES capacity configuration and scheduling results, the cooling, heating and power supply can satisfy the demand of the data center while generating over 30% of the power from renewable sources. In addition, carbon tax is considered in the optimization model, which accounts for 4% of the life-cycle cost. Furthermore, the sensitivity of carbon tax, equipment lifespan and predicted mean vote is analyzed, obtaining that these factors have some influence on the performance of the system. Finally, the uncertainty impact on system planning is studied, concluding that the designed system exhibits high robustness levels when renewable energy forecast error is less than 20% or load forecast error is less than 10%. • A DMES for data centers includes wind, solar and geothermal energy is established. • A life-cycle-based multi-objective optimization design framework for DMES is built. • An improved multi-objective grasshopper optimization algorithm is proposed. • Renewable energy power generation over 30%, while carbon taxes made up 4% of LCC. • Design has high robustness when renewable energy error <10% or power load error <20%. [ABSTRACT FROM AUTHOR]

Data centers (DCs) uninterruptedly run 24/24 h, 365 days per year with much huge operating scale, and have the characteristics of high operation safety requirement, high heat flux density, high energy consumption and high carbon emission. They are influential energy consumers and carbon emitters in building or even global energy sectors (around 3% of global energy consumption), who are also significant waste heat producer (e.g., waste heat from year-round uninterrupted operation of IT equipment and cooling system). Huge energy consumption has increased the burden on the global energy industry, while carbon and direct waste heat emissions have also caused great damages to the outdoor environment. Thus, it is critical to improve the energy efficiency in DCs and to realize the energy conservations and environmental deterioration alleviation. Waste heat recovery technology is considered as a promising approach to improve energy efficiency, achieve energy and energy cost savings, and mitigate environmental impacts (caused by both carbon emission and waste heat discharge) at the same time. This article conducts a comprehensive review on recovering waste heat from all kind of sources (e.g., exhaust air, circulating water, and coolants) in DCs for various energy uses (e.g., heating supply, district heating supplement, cooling and electricity productions, and industrial/agricultural production process) and different application scenarios (e.g., office buildings, comprehensive energy community and residential buildings), while the future research and development proposals for DC waste heat recoveries are given through technical, energy, environmental and economic analysis. • Conduct a comprehensive review on waste heat recoveries in data centers • Consider various waste heat sources (exhaust air, circulating water, and coolants) • Regard various waste heat uses (heating, DH, cooling and power generation, etc.) • Regard various applications (office, residential buildings, and energy community) • Give recommendations for future study through technical and 3 E analysis [ABSTRACT FROM AUTHOR]

The number of data centers worldwide has been growing exponentially to serve a multitude of aspects of modern life. The major challenges that face data centers operators are the rising electricity bills, growing carbon footprints, and unexpected power outages. An effective approach to tackle such problems is to investigate the implementation of efficient power supplies that offers the rack voltage levels required to power servers. Recently, the differential power processing (DPP) concept proved to potentially offer less number of cascaded stepping down stages by connecting the servers in series and introducing a DPP converter to supply or withdraw current mismatch in the series connection. DPP system can replace the conventional two-stage power electronics converter with an efficiently designed converter. Thus, the series stacking power delivery architecture is proposed with efficient Gallium Nitride (GaN) based power electronics converters to reduce the system size and losses. The introduced review in this paper aims to give the reader a comprehensive understanding of DPP implementation for data centers in terms of the possible series stacking architecture, converter types and switch types. The advantages and disadvantages of each architecture are discussed along with the most commonly employed DPP converters in various applications. Further investigation on the cost-effectiveness and commercialization of GaN-based series stacking architecture still needs further investigation and deployment. [ABSTRACT FROM AUTHOR]

Dynamic performance of a 10-kW proton exchange membrane fuel cell (PEMFC)–battery hybrid system to power servers in data centers has been experimentally evaluated in our previous work [1]. The present work is a numerical study based on the previous work to identify the dynamic characteristics and present basic insights for the system control strategy during transients. The hybrid system dynamic model has been developed using the MATLAB–Simulink®, which consists of a one-dimensional, two-phase dynamic model of the PEMFC, lumped dynamic model of an air blower and a battery. The system model is verified by comparing the dynamic behavior of the power generated by the PEMFC and battery with the experimental data at the step change of the system demand power between 0 and 1.5, 3.0, 4.5, 6.0, 7.5, and 9.0 kW. During the step load increases, the system instantly obtained the total amount of external load power from the battery within 0.1 s in every case and gradually decreased until approximately 4 s or 6 s as the power generated by the fuel cell is gradually increased. The dynamic response of the system model is compared with the experimental data at various load profiles of three, six, and nine servers. • Developing a dynamic model of a PEMFC-battery hybrid system for powering servers. • Comparing the PEMFC stack model with experiments at the steady-state. • Verification of the PEMFC hybrid system model at the step changes of load power. • Verification of the PEMFC hybrid system model at the server load power profiles. • Current model can be used to develop the optimal control strategy of hybrid system. [ABSTRACT FROM AUTHOR]

In the past two decades, data centers have significantly grown in size and number around the globe, leading to substantial energy and water stress wherever they operate. Thermal management typically represents half of the total energy consumption in traditional data centers. This work studies the energy and water savings that air-side free-cooling yields under diverse weather conditions. We modeled a free-cooling system and a conventional data center using MATLAB. The hourly weather data was obtained from a publicly available online repository for the Chilean territory. Using air-side free-cooling for data center thermal management is feasible in various weather conditions. Locations with colder temperatures (< 20 °C) and higher humidity (> 60%) are more favorable as they require minimal use of the chiller and/or humidifier. These conditions are present in BSk'(s) Cold Semi-arid climate with dry summer and oceanic influence, as well as Csb' Mediterranean climate (warm summer) with oceanic influence. However, locations with drier air (< 50%) and warmer temperatures (> 20 °C), such as those found in ET(w) Tundra climate with dry winter, BWk(w) Cold Desert climate with dry winter, or BSk(s) Cold Semi-arid climate with dry summer, are less suitable for air free cooling. Even under suboptimal energy and water savings, the air-side economizer still performed considerably better than a conventional data center cooling method (using CRAH units). • Free-cooling and conventional data centers are modeled using hourly weather data. • Air-side free-cooling economizer may save energy and water depending on the weather. • BSk'(s) and Csb' climates show significant energy and water savings. • Negligible usage of chiller and humidifier in some weather conditions. [ABSTRACT FROM AUTHOR]

Exponential growth of traffic and bandwidth demands in current data center networks requires low-latency high-throughput interconnection networks, considering power consumption. By considering growth of both multicast and unicast applications, power efficient communication becomes one of the main design challenges in today's data center networks. Addressing these demands, optical networks suggest several benefits as well as circumventing most disadvantages of electrical networks. In this paper, we propose an all-optical scalable architecture, named as OMUX, for communicating intra-data centers. This architecture utilizes passive optical devices and enables optical circuit switching without any switch configuration or path reservation. In addition, it is customized for both unicast and multicast transmissions. It can replace aggregation and core switches in the network, to reduce power consumptions. Simulation results verify its lower latency, higher throughput, and reduced power consumption, compared to conventional electrical or alternative optical solutions. [ABSTRACT FROM AUTHOR]

Abstract A recent research showed that data centers consumed about 1.3% of total worldwide electricity. Thus, energy conservation in data centers should be paid enough attention to. The impacts of server lower-side terminal baffles (SLTBs) on the thermal environment in data center are optimized by experiments and simulations. In this paper, three different sizes (4 cm × 46 cm, 8 cm × 46 cm and 12 cm × 46 cm) of baffles with six different baffle angles (0°, 15°, 30°, 45°, 60° and 75°) are applied to investigate the optimum airflow distribution. The simulation model was validated by on-site experiments, while the experimental results are in good agreement with numerical results. The measurements show that the use of 45° angle and 8 cm × 46 cm SLTBs can effectively eliminate the heat accumulation and improve the airflow distribution and cooling efficiency in data center. In addition, it can decrease the temperature of the rack hotspot 1.9–2.5K, that reduces significantly the risk of local hotspot and avoid catastrophic damages to the servers. Furthermore, the exhaust air temperature distribution of the upper and lower parts of rack was quite uniform and by using the optimal baffles, the maximum temperature difference reduced from 2.2K to 0.5K. Highlights • An innovative airflow management method in data center is applied in this paper. • Flexible baffles are set on the lower side of servers' rear terminal to improve thermal environment. • The optimum thermal distribution appears in data center with 45° and 8 cm server lower-side baffles. • The exhaust air temperature distribution of the upper and lower parts of the rack was relatively uniform. • The temperature drop of rack hotspot reaches 1.9–2.5 °C reducing the risk of local hotspot damages to the servers. [ABSTRACT FROM AUTHOR]

Abstract Social media have emerged in the last decade as a viable and ubiquitous means of communication. The ease of user content generation within these platforms, e.g. check-in information, multimedia data, etc., along with the proliferation of Global Positioning System (GPS)-enabled, always-connected capture devices lead to data streams of unprecedented amount and a radical change in information sharing. Social data streams raise a variety of practical challenges, including derivation of real-time meaningful insights from effectively gathered social information, as well as a paradigm shift for content distribution with the leverage of contextual data associated with user preferences, geographical characteristics and devices in general. In this article we present a comprehensive survey that outlines the state-of-the-art situation and organizes challenges concerning social media streams and the infrastructure of the data centres supporting the efficient access to data streams in terms of content distribution, data diffusion, data replication, energy efficiency and network infrastructure. We systematize the existing literature and proceed to identify and analyse the main research points and industrial efforts in the area as far as modelling, simulation and performance evaluation are concerned. [ABSTRACT FROM AUTHOR]

Data centers are considered a critical form of infrastructure in today's information society. They are at the core of the cloud computing and services revolution, which has changed the business models on how organizations deal with IT infrastructure costs. The hybrid data center architecture incorporates both legacy and fully virtualized infrastructures. On the one hand, the composite infrastructure has improved resource utilization and consolidation by adding flexibility and scalability factors, making the data center more cost effective and more agile. On the other hand, the hybrid infrastructure has imposed a new set of security challenges that need to be brought into focus. The lack of resource availability can be a great risk for delay sensitive applications such as voice over IP (VoIP) and online gaming when cloud computing is the deployment model. This study addresses the emerging risk problem by conducting a comprehensive security risk assessment using the NIST national vulnerability database (NVD) combined with EBIOS risk analysis and evaluation methodology. This study focuses on resource availability problem emanating from delay variations and queuing mechanisms in virtualized systems and its impact on delay sensitive applications. The study argues for the existence of availability risk within the hybrid data center infrastructure, which can deteriorate the performance of delay sensitive applications. Security remedial and countermeasures to the identified security risks are suggested in an extended discussion at the end of the study. [ABSTRACT FROM AUTHOR]

Rural towns across the US have undergone economic and social restructuring. We focus on two former timber towns adjacent to public lands that are transitioning to new production economies, wherein the communities develop new economic sectors. One of our cases has developed data centers, and the other prisons. Through in-depth interviews with residents, we explore the shifts toward these new sectors and their impacts on community social adaptability, in particular human and social capital and community identity. In both of the communities, there are well-paying jobs in the new production sectors. However, indicators of social adaptability have diverged greatly. In the data center community, there are higher perceived levels of human and social capital, while the prison community has lower levels. Members of both communities describe their community identity as in flux, and voice apprehension about growth and development of new industries, including the possibility that the communities may be shifting single-industry dependence from one sector (timber) to a new sector. Both communities are rebuilding their relationships to adjacent public lands and view those public lands as opportunities for economic diversification. • Communities adjacent to public lands are undergoing socioeconomic transitions • We examine two new production transitions through case studies • We center our two cases on understudied economies: data centers and prisons • The transitions exhibit different impacts on social adaptability indicators • Both communities retain (and have re-configured) connections to nearby public lands [ABSTRACT FROM AUTHOR]

The high power of blade servers inside small data centers (DCs) can cause heat accumulation, which can degrade the performance of DCs. To address the problem of heat dissipation, we performed periodic experiments on a thermal management system based on a phase change material. The effects of time ratio, heating power, and cooling water temperature on the performance of thermal control and periodic stability were investigated. Information entropy reflects uncertainty by transforming a variable changes over time into a single parameter representing inhomogeneity of time. Herein, it was introduced as an evaluation index of thermal control performance. The results verified that the peak temperature of the heating surface can be maintained below 80 °C for several periods under specific conditions. Additionally, the information entropy results of the experimental data reflected the thermal control performance of the system and validated the experimental conditions. Based on the experimental results and the corresponding information entropy analysis, we proposed a design strategy for the system parameters and a method for reducing the number of measuring points. The obtained results formed the basis for the thermal management design of the server level of DCs, other electronic equipment, and electric vehicles. • Periodic performance of a PCM thermal management system is experimentally analyzed. • Information entropy evaluates the effects of system parameters on thermal control. • Design strategy and method for reducing number of measuring points are proposed. • Complex data are integrated and the periodic performance analysis is simplified. • Results can aid in designing effective thermal management systems for data centers. [ABSTRACT FROM AUTHOR]

The main purpose of this paper is to research energy efficiency by using free cooling in data centers. The upper limit of wet bulb temperature of free cooling is obtained and the recommended value is 16 ℃ in the way of indirect cooling by open cooling tower. The concept of sensitivity of the natural cold resource is proposed for the first time. The natural resource utilizing potential of main climate zones of China is analyzed. The analysis indicates that natural resource utilizing ratio is over 50% in most regions. The improvement of the upper limit of wet bulb temperature is most beneficial to Moderate area. Finally, a data center in hot-summer and warm-winter region is analyzed. The results show that the annual synthesized COP value increased by 23.7% and the energy saving rate is about 19.2% after full use of free cooling. [ABSTRACT FROM AUTHOR]

Data centers have the characteristics of high heat production from equipment and uneven temperature distribution. Thus, in order to improve the thermal environment in data centers, this paper puts forward an air-conditioning system with under-floor integrated solar chimney. The improved system can also reduce energy consumption due to annual cooling. A typical data center model is built and verified with experiments. Based on this model, the new model integrated solar chimney is built and simulated. The results show that after integrating the solar chimney, the heat island effect impairs and the temperature of all compared points have a decline, which reached an average decline of about 4.4 ℃. In addition, integrated solar chimney system improves the mixing of cold air and hot air effectively. [ABSTRACT FROM AUTHOR]

Airflow pattern in a data center impacts the cooling effectiveness of the air-conditioning systems significantly, and hence impacts the energy efficiency of the whole data center. This paper presented investigation of 3 airflow patterns for a typical data center by CFD simulation. An under-floor air supplying system, row-level cooling system and rack level cooling system was used in the targeted data center. Cooling performance was simulated by CFD software. The cooling performance was compared by indices like Index of Mixing (IOM), Return Temperature Index (RTI), Temperature Inequality Index K T , Mean Temperature Difference Index of Rack Intake and Exhaust Air, Energy Utilization Coefficient. The results indicated that rack level airflow pattern shows the best performance in data center cooling. Compared with under-floor air supplying system, the IOM index is only 0.0011, and the K T index is 0.0082. In the end an optimization method was proposed for the deficiencies in rack level cooling system. [ABSTRACT FROM AUTHOR]

Cloud computing and web applications have further highlighted the need for powerful data centers with large bandwidth and low power consumption. In order to provide such a large bandwidth, today's electronic data centers require high power levels. Furthermore, communications and devices are not always used to their full potential. Optical networks can be a proper option for data centers, as they can offer large and variable bandwidth with lower power consumption compared to their electronic counterparts. In this regard, a dynamic all-optical network architecture is proposed in this paper, with bandwidth reconfiguration capability and low latency to reduce energy consumption. This proposed architecture, called Flat Ball, offers dynamic bandwidth utilizing passive optical devices with low power consumption and latency. Under realistic data center traffic scenarios, the latency of this network is up to 50% lower than its electrical counterpart, while providing a much higher throughput and consuming up to 60% less power. [ABSTRACT FROM AUTHOR]

In present computer structures in view of distributed data centers (DC) provides flexible management of network topology and functions in real time. This method allows to get rid of the excess costs of expensive equipment and network service. To provide resiliency of network infrastructure of DC it is necessary to use effective fast rerouting algorithms. In this work we propose adaptive rerouting algorithm of data flows in distributed DC based on the paired shifts data. [ABSTRACT FROM AUTHOR]

Generalized few-mode-fiber (FMF) transmission uses N coherent receivers for mode detection, where N scales with the number of fiber modes. Multiple coherent receivers increase the cost of optical network units (ONUs) in access networks, specifically for intra- and inter data center applications. We have experimentally evaluated a cost effective low-complexity receiver architecture based on mode-division multiplexing (MDM) to mode frequency-division multiplexing (MFDM) conversion for data center connectivity. A single coherent receiver is used to efficiently detect 10 Gbit/s QPSK Nyquist signals over 2.8 km 4-LP mode graded index fiber reducing the ONUs complexity to N / 4 . All the transmitted modes are successfully detected below the forward-error-correction (FEC) limit, i.e. 1×10 −3 BER. [ABSTRACT FROM AUTHOR]

Data centers are amongst the most complex and energy demanding building environments. The trends of increasing data center capacity and power use by information technology equipment, re-emphasize the need to reduce cooling loads, first cost for HVAC equipment, and finally related energy and operational costs. The first part of the paper presents an overview of relevant information and available resources for the efficient design and operation of these critical environments. The second part considers two large data centers located in Athens that serve two major Hellenic banks as case studies in order to investigate and quantify the potential measures for improving their energy performance. The two facilities were audited and actual energy consumption were used for an energy assessment. According to simulation results of various efficiency measures, the Power Usage Effectiveness may be improved from 2.49 to 2.11 for the first case study and from 1.89 to 1.66 for the second one. [ABSTRACT FROM AUTHOR]

In recent years, the energy consumption of data centers (DCs) has increased rapidly. Since the demand response (DR) capability of DCs is considerable, DCs play an important role in DR programs. In this work, we further focus on the DCs' integrated demand response (IDR) capability which owes to its temporally and spatially shiftable electricity and heat. To make full use of waste heat retained in the coolant, we design a waste heat recovery process for waste heat reutilization. Moreover, we propose a decentralized optimization algorithm based on the alternating direction multiplier method (ADMM) to protect the inner sensitive data of DCs. Numerical results show that energy purchase costs can be effectively reduced for DCs with IDR capability. It is also proved that the reutilized waste heat constitutes an important part of the heat supply. Besides, the convergence performance of the proposed algorithm is verified based on large quantities of simulations. • Design a waste heat recovery process for waste heat reutilization in data centers. • Construct an integrated demand response model of data centers to reduce daily costs. • Propose an ADMM-based decentralized optimization algorithm to protect the inner sensitive data of data centers. [ABSTRACT FROM AUTHOR]

Data Centers are huge power consumers, both because of the energy required for computation and the cooling needed to keep servers below thermal redlining. The most common technique to minimize cooling costs is increasing data room temperature. However, to avoid reliability issues, and to enhance energy efficiency, there is a need to predict the temperature attained by servers under variable cooling setups. Due to the complex thermal dynamics of data rooms, accurate runtime data center temperature prediction has remained as an important challenge. By using Grammatical Evolution techniques, this paper presents a methodology for the generation of temperature models for data centers and the runtime prediction of CPU and inlet temperature under variable cooling setups. As opposed to time costly Computational Fluid Dynamics techniques, our models do not need specific knowledge about the problem, can be used in arbitrary data centers, re-trained if conditions change and have negligible overhead during runtime prediction. Our models have been trained and tested by using traces from real Data Center scenarios. Our results show how we can fully predict the temperature of the servers in a data rooms, with prediction errors below 2 °C and 0.5 °C in CPU and server inlet temperature respectively. [ABSTRACT FROM AUTHOR]

The exponential data growth and demands for high computing resources lead to excessive resource use in cloud data centers, which cause an increase in energy consumption and high carbon emissions in the environment. So, the high energy consumption, inefficient resource usage, and quality of service assurance (QoS) are major challenges for cloud data centers. The dynamic consolidation of Virtual Machines (VMs) is proven to be an efficient way to tackle these issues while reducing energy consumption and improving resource utilization in data centers. It reduces the number of active hosts for energy efficiency by switching under-utilized and idle hosts into lower power mode. So, several heuristics and Artificial Intelligence (AI) based VM consolidation approaches have been published in papers. Most existing approaches rely on the aggressive consolidation of VMs for energy efficiency, thus causing performance degradation and high SLA violation. However, an automated solution is needed to reduce energy consumption and SLA violation by ensuring efficient resource usage in the cloud data center environment. Therefore, this article proposes an energy-efficient autonomous VM consolidation (AVMC) mechanism that has Deep Reinforcement Learning (DRL) based agent for performing VM consolidation decisions. The DRL agent learns the optimal distribution of VMs in the data center, considering energy efficiency and QoS assurance. The real-time workload traces from PlanetLab have been used to validate the proposed mechanism. Experimental results reveal the superiority of the proposed AVMC system over the existing models. AVMC reduced the energy consumption and SLA violation rate significantly. • DRL-based Autonomous VM consolidation (AVMC) mechanism is presented for cloud data centers. • The main goal is to maximize energy saving and minimize SLA violations. • Real workload traces have been used for the experimental evaluation. • Several performance metrics have been used to compare the AVMC method with existing approaches. [ABSTRACT FROM AUTHOR]

• Six combined cooling and power systems with multi-mode operation are proposed. • The proposed systems can realize diversified power and cooling supply. • Evaluated by parametric study and multi-objective optimization. • The ORC/ARC is considered as the optimal system in any of the operation mode. • The energy efficiency at the optima is 58.13% and unit cost of product is 4.25 $/GJ. Current solutions for waste heat recovery from data centers are mainly district heating and refrigeration. However, intermittent demand for space heating and cooling resulted from seasonal or spatial variations makes inefficient utilization of waste heat. To better harvest the low-grade thermal energy in data centers, combined cooling and power systems are proposed, with the capability of adjusting the energy output to meet diverse energy demands hourly and seasonally. In total, six cogeneration systems with different configurations are investigated, with the power cycles being either organic Rankine cycle or Kalina cycle, and the refrigeration cycles being vapor compression refrigeration cycle, ejector expansion refrigeration cycle or absorption refrigeration cycle. Depending on the energy demands of data centers, each system has three operation modes: combined cooling and power, power alone and cooling alone. The performance is evaluated in terms of both the thermodynamic and economic performance by parametric study and multi-objective optimization. Results show that the organic Rankine cycle/absorption refrigeration cycle hybrid system has the best thermal-economic performance in all three modes. According to the comprehensive evaluation of the TOPSIS method and the entropy weight method, the energy efficiency at the optima is 58.13%, and the unit cost of product is 4.25 $/GJ, with the net power output of 4.25 kW and the cooling capacity of 150 kW. [ABSTRACT FROM AUTHOR]

• A thermodynamically-rigorous model enables reliable PUE and WUE estimations. • A PUE variance of 62% is quantified, depending on technology, efficiency, climate. • A WUE variance of 100% is quantified, depending on technology, efficiency, climate. • Best-practice values and their determinants are quantified. • Results enable more reliable DC direct and indirect water footprint estimation. The water use of data centers (DCs) is becoming an increasingly important consideration for sustainability analysts, but has been difficult to quantify due to lack of reported data by DC operators. This work develops thermodynamically-compatible water use effectiveness (WUE) and power usage effectiveness (PUE) ranges needed for direct and indirect water use analysis for 10 different DC archetypes in 15 U.S. climate zones, using a hybrid physical-statistical approach that is validated with real-world data. The ranges further capture variabilities between best and poor DC efficiency practices, showing that PUE and WUE values can have relative difference as much as 62% and 100% respectively, depending on cooling technologies, efficiencies, and locations. PUE results indicate significant climate effects, whereas WUE results indicate the strongest climate effects for airside economizers with either adiabatic or water-cooled chiller systems. Collectively, the results underscore the importance of considering technology, efficiency, and climate effects when estimating PUE and WUE for direct and indirect DC water use analysis. Results also identify variables leading to best-achievable PUE and WUE values by climate zone and cooling system type—including operational set points, use of free cooling, and cooling tower equipment and operational factors—which can support DC water- and energy-efficiency policy initiatives. [ABSTRACT FROM AUTHOR]

The rapid development of digital economy has led to a dramatic increasement of internet data centers (IDCs), which consume a large amount of electricity. The diverse data-processing demands, high power consumption and workload uncertainty put forward a high requirement for the economical and secure operation of IDCs. As information technology (IT) devices are driven by direct current (DC), flexible substation (FS) has been gradually utilized to provide DC power for IDCs. This paper proposes robust operation strategies for minimizing IDC power consumption with FS integration. First, the linearized IDC power consumption model based on the technology of dynamic voltage and frequency scaling (DVFS) is proposed to describe the operating state of IT devices. Then, considering the FS-based coordinated operation of IDC, photovoltaic (PV) station and energy storage system (ESS), the deterministic power consumption minimization model of IDCs is established. Considering the workload uncertainty, the operation strategies based on distributionally robust optimization (DRO) for IDCs are further proposed. Finally, the effectiveness of proposed method is validated on a modified practical network with IDCs integrated. Results show that through the dispatch of workload and power flow regulation of FS, the power consumption and dropping workload of IDCs are effectively reduced. • The linearized power consumption model of data centers is proposed. • Variable resources are coordinated with flexible substation integration. • Distributionally robust optimization strategies are proposed for data centers. • The power consumption and dropping workload are effectively reduced. [ABSTRACT FROM AUTHOR]

• Two DCs located in a hot-arid climate with distinctive cooling systems are compared. • A novel method is presented for water-energy usage trade-off in DCs. • Synergistic water and energy conservation approaches are presented. Although energy and water are intertwined, the analysis of energy consumption in data centers (DCs) has largely ignored water needs. In this study, we utilize established DC efficiency metrics, power usage effectiveness (PUE) and water usage effectiveness (WUE), to analyze the water and energy use in two colocation DCs (DC1 and DC2) located in the greater Phoenix area. These DCs use distinct cooling systems: DC1 uses air-cooled chillers and DC2 uses both water-cooled chillers and evaporative cooling. We also present a method for analyzing the water-energy tradeoff in DCs as a useful tool for DC designers and managers to evaluate different cooling systems. The results highlight that although DC1's average PUE is nearly 13% higher compared to DC2's average PUE, its source WUE is 66% lower than DC2. In addition, DC1's PUE is more affected by seasonal changes (variation of dry-bulb temperature) , while DC2's WUE has a larger seasonal variation. Our cooling system analysis indicates that hybrid evaporative cooling has the least power consumption, while air-cooled chillers have the most water use. Utilizing free cooling and evaporative cooling when outside air condition permits (∼40% per year), results in substantial power and water savings in Phoenix DCs. [ABSTRACT FROM AUTHOR]

• A green and sustainable method for packing servers is investigated. • Cooling principles on cooling efficiency and packing capacity are constructed. • The problem is formulated as bin-packing models via segmenting and gridding. • Mathematical analyses and solution strategies help to pack servers efficiently. An Underwater Data Center (UDC) is an underwater vessel full of computing servers and designed with a cooling system using cold water in the ocean. A UDC vessel is composed of cabinets for computing servers, and the cabinets are finally packed into racks that facilitate the installments of the computing servers. We formulate the problem of packing cabinets into the vessel by considering the water volume required for server cooling and the placement of the cabinets of servers. Due to different patterns of racks, a set of mixed-integer linear programs were developed, and they were solved by methods with several stages. As demonstrated by experiments, the proposed models and strategies effectively solve the simulated instances generated by considering the cooling principles. The cabinets with more considerable differences in their width and height are better than single-sized cabinets because the smaller cabinets make the packing process more flexible. Moreover, multi-sized cabinet packing is significantly better than single-sized cabinet packing. Additionally, it is not always appropriate to use racks with large capacities. However, racks with multiple capacities may improve the space and energy utilization degrees when the size and the designed heat power consumption of the data center are fixed. [ABSTRACT FROM AUTHOR]

With the advantages such as high-performance, low-maintenance, and reliability, more and more companies are moving their computing infrastructures to the cloud. In the meantime, with the increasing number of users continuously submitting jobs to cloud, energy consumed by the current cloud data centers has become a major concern for cloud service providers, due to financial and environmental reasons. In this paper, we propose a deep reinforcement learning (DRL) approach to handle real-time jobs. Specifically, we focus on allocating incoming jobs to appropriate virtual machines (VMs) in a way that energy consumption can be optimized while high quality of service (QoS) can be achieved. We give the detailed design and implementation of our approach, and our experimental results demonstrate that the proposed method can achieve better performance in job success rate and average response time with less energy consumption than the current approaches, in the presence of different real-time cloud workloads. [ABSTRACT FROM AUTHOR]

• Integrating power-to-hydrogen, electric vehicles and data center under in MEM. • Proposing robust-stochastic strategy to control decision-making in presence of various uncertainties. • Establishing a multi-objective robust-stochastic strategy for the MEM. • Electricity flow, heating flow, and cooling flow are provided to the data centers and adjacent commercial and residential buildings. To achieve zero net emission conditions, multi-energy microgrids (MEMs) have grown rapidly in recent years. Hydrogen-based technologies and energy conversion systems bring unique opportunities for decreasing carbon emissions in MEMs. Hence, this paper tries to optimize the scheduling of the integrated energy sources in a MEM coupled with promoted energy conversion facilities and the hybrid demand response (DR) scheme. To this end, a hybrid robust-stochastic methodology is established in the form of the multi-objective optimization problem to minimize the operation cost and emission. The developed energy management model for MEM is used to serve and manage the behavior of the data centers and adjacent buildings as the flexible/non-flexible demands. The proposed strategy provides a continuous control mechanism for MEM's operator in the presence of various energy suppliers like hydrogen-based technologies, i.e., power-to-hydrogen unit, hydrogen storage system (HSS), and fuel cell, battery electric vehicle (BEV), etc. Herein, the uncertain characteristics of the electricity market price, energy demands, and renewable power production are controlled by the robust approach and several scenarios with appropriate probabilities. The proposed strategy is applied to various case studies of a typical MEM associated with data centers in the day-ahead energy markets. According to the obtained simulation results, the proposed energy management strategy for MEM can reduce the operation cost and emission by up to 8.2% and 3.9%, respectively. [ABSTRACT FROM AUTHOR]

Abstract: Modern data centers must provide performance assurance for complex system software such as multi-tier web applications. In addition, the power consumption of data centers needs to be minimized to reduce operating costs and avoid system overheating. Various power-efficient performance management strategies have been proposed based on dynamic voltage and frequency scaling (DVFS). Virtualization technologies have also made it possible to consolidate multiple virtual machines (VMs) onto a smaller number of active physical servers for even greater power savings, but at the cost of a higher overhead. This paper proposes a performance-controlled power optimization solution for virtualized server clusters with multi-tier applications. While most existing work relies on either DVFS or server consolidation in a separate manner, our solution utilizes both strategies for maximized power savings by integrating feedback control with optimization strategies. At the application level, a novel multi-input–multi-output controller is designed to achieve the desired performance for applications spanning multiple VMs, on a short time scale, by reallocating the CPU resources and conducting DVFS. At the cluster level, a power optimizer is proposed to incrementally consolidate VMs onto the most power-efficient servers on a longer time scale. Empirical results on a hardware testbed demonstrate that our solution outperforms pMapper, a state-of-the-art server consolidation algorithm, by having greater power savings and smaller consolidation overheads while achieving the required application performance. Extensive simulation results, based on a trace file of 5415 real servers, demonstrate the efficacy of our solution in large-scale data centers. [Copyright &y& Elsevier]

Abstract: In the recent years, energy-efficiency of computing infrastructures has gained a great attention. For this reason, proper estimation and evaluation of energy that is required to execute data center workloads became an important research problem. In this paper we present a Data Center Workload and Resource Management Simulator (DCworms) which enables modeling and simulation of computing infrastructures to estimate their performance, energy consumption, and energy-efficiency metrics for diverse workloads and management policies. We discuss methods of power usage modeling available in the simulator. To this end, we compare results of simulations to measurements of real servers. To demonstrate DCworms capabilities we evaluate impact of several resource management policies on overall energy-efficiency of specific workloads executed on heterogeneous resources. [Copyright &y& Elsevier]

Data centers are power hungry facilities. Energy-aware task scheduling approaches are of utmost importance to improve energy savings in data centers, although they need to know beforehand the energy consumption of the applications that will run in the servers. This is usually done through a full profiling of the applications, which is not feasible in long-running application scenarios due to the long execution times. In the present work we use an application signature that allows to estimate the energy without the need to execute the application completely. We use different scheduling approaches together with the information of the application signature to improve the makespan of the scheduling process and therefore improve the energy savings in data centers. We evaluate the accuracy of using the application signature by means of comparing against an oracle method obtaining an error below 1.5%, and Compression Ratios around 39.7 to 45.8. [Display omitted] • Energy-aware task scheduling approaches reduce energy consumption in data centers. • They rely on a full profiling, not feasible in long-running applications. • The application signature is used to avoid the full profiling. • The application signature makes the task scheduling approaches more efficient. [ABSTRACT FROM AUTHOR]

• A framework for battery expansion planning is proposed in distribution network. • The framework captures the interactions among distribution and data network operators. • The uncertainties in demand, workload, PV generation, and line outages are considered. • DS3 algorithm is used to determine the representative scenarios. This paper presents a stochastic expansion planning framework to determine the installation time, location, and capacity of battery energy storage systems in the distribution networks with considerable penetration of photovoltaic generation and data centers. The presented framework aims to minimize the capital cost of the battery energy storage and the operation cost of the distribution network while ensuring the security of energy supply for the data centers that serve end-users in the data network as well as the reliability requirements of the distribution network. The proposed stochastic framework captures the interactions between the distribution network and data center operators considering limited shared information among these entities. Benders decomposition is used to capture the interactions between these autonomous operators in the electricity and data networks. The uncertainties associated with the electric demand, data center workload, solar PV generation, and the availability of the distribution branches are captured using Monte Carlo simulation. The representative scenarios are selected using a dissimilarity-based sparse subset selection algorithm. To evaluate the effectiveness of the proposed framework, numerical results are presented for a modified IEEE-34 bus distribution network with data centers and PV generation units. [ABSTRACT FROM AUTHOR]

• A ren-aware scheduler is proposed that schedules jobs on data centers with that have both green and brown sources of energy available to them. • The objective of the ren-aware scheduler is to increase the "ren-percentage" or the fraction of the energy used that is derived from renewable sources. • We consider real data from computing traces, as well as solar irradiation data from different geographic locations. • Our results show that there exists a trade-off between ren-percentage and the average waiting time of jobs. • We also suggest a method to use data centers in geographically dispersed locations to improve the ren-percent. The rise in the penetration of the internet across the world has led to a rapid increase in the consumption of energy at the data centers established by leading cloud data service providers. High power consumption by these data centers [DCs] leads to high operational costs and high carbon emissions into the environment. From a sustainability point of view, the ultimate goal is to maximize the productivity and efficiency of these data centers while keeping greenhouse gas emissions to the minimum and maximize data center productivity. This goal can be achieved by better resource utilization and replacing carbon-intensive approaches of energy production with green sources of energy. Due to the limited intermittent availability of renewable sources of energy, the ideal 'Green' design for the DCs, should incorporate inter-operability with both renewable and non-renewable sources of energy. In this paper, we propose a ren-aware scheduler to schedule computational workload by prioritizing their execution within the duration of green energy availability on the basis of the predicted hourly green energy and workload data of DCs. Our results demonstrate that our ren-aware scheduler can increase the green energy consumption by 51% compared to the conventional randomized scheduler that distributes load without considering green energy and load. It can also reduce the total energy consumption by 25% by putting the DCs to sleep during their idle time, as it saves 4.5 times more idle energy than the randomized scheduler. Additionally, the results also demonstrate how the role of time zones of the DCs and the duration of green energy availability in them is pivotal in our ren-aware scheduler's performance. [ABSTRACT FROM AUTHOR]

Abstract: Data centers represent today's major energy consumers. This is especially true in the Netherlands, informally called “the data center country” for Amsterdam includes 33% of the data centers of the whole European Union. To address this problem, the Dutch Government initiated the MJA (Meerjarenafspraken, or in English multiannual agreements), a nationwide initiative intended to improve energy efficiency of ICT products, services and processes. Major energy consumers, like data centers, actively participate in the MJA initiative to lower and optimize their energy consumption. To this end, the MJA members have documented the green ICT practices developed so far. The resulting documentation consists of an excel sheet of 83 consolidated practices. As a way to formalize, classify and share green practices, in our previous work we have proposed a green strategy model and have mapped the MJA practices accordingly. In this way we made explicit the fundamental knowledge elements of a green practice, namely the MJA green goals, green practices, and practice's effects. In this work, we take a further step into the content of the codified practices, by: i) classifying the practices based on the concerns they address, ii) looking into the goals that data centers intend to achieve in relation to the concerns that practices address, iii) classifying the practices based on the role that IT plays, and iv) distinguishing practices that are software-related from non-software related ones. In this way, we provide a picture of how data centers currently address the problem of reducing their energy footprint, possibly offering inspiration to both researchers (in progressing in their research further from what already exists), and practitioners (in reusing practices and learning what knowledge elements are important for informed decision making). [Copyright &y& Elsevier]

• Control and energy management systems are proposed for a green data center on a remote island. • Hybrid energy storage system is used. • Real data of the Alderney Race tidal current speed and Cherbourg solar radiation are used. Data centers consume a large amount of energy, contributing to CO 2 emissions, global warming, and resulting in significant electricity cost. To address these concerns, an increasing number of companies have considered building green data centers. The most abundant energy resource is solar; it is now a key player in the global energy transformation. Tidal energy has also received particular attention recently. The predictable characteristic of the resource makes the kinetic energy of tidal current an extremely competitive power resource compared to other commonly used renewable energies. For this purpose, a hybrid Tidal/Photovoltaic system is used for powering an MW scale green data center on a remote island. Green data centers are mainly dependent on renewable energies, which have intermittent nature and require a storage system capable of ensuring sustainable energy feeding. The proposed system consists of an MW scale proton exchange membrane electrolyzer and fuel cell. This system is associated with the LiFePO 4 battery to cover the fast-electrical dynamics. In this paper, the system modeling is presented along with an initial proposal of control and energy management system to align the data center energy consumption with the renewable energy generated while respecting the different system constraints. The model is implemented in Matlab/Simulink platform where the simulation results exhibit the system performance under different operation conditions. [ABSTRACT FROM AUTHOR]

Problems of hot air recirculation, hot spots, and bad performance indices of data centers at terminal racks were noticed and reported in the literature. The present work investigates the possible management solutions of heterogeneous thermal problems for terminal racks in data centers in case of using in-row cooling racks arrays. Three possible solutions, namely using top aisle containments, top and side aisles containments, and air conditioning unit at the start and end of each rack row, are investigated. Computational Fluid Dynamics(CFD) using ANSYS-Icepak software as a specialized software in cooling electronic devices is used to conduct this study. A prototype of a small-scale data center module is experimentally constructed and tested to validate the present CFD work. Supply Heat Index (SHI), Index of Mixing (IOM), Beta Index (β) and Energy Utilization Index (η r) are utilized to investigate, assess and compare the different suggested solutions. The results show that (i) problems of hot air recirculation, hot spots, and bad performances indices are noticed for the terminal racks if no solution is applied, (ii) using top aisle containments partially solve the heterogeneous thermal problem of terminal racks, (iii) using top and sides containments completely solve the problem of hot air recirculation (iv) starting and ending the racks row with computer air conditioning units solve the problem of the heterogeneous distribution of intake air temperature along with the racks height and (v) Energy Utilization Coefficient of terminal racks is increased by 45% if the top and sides containments are applied or starting the racks row with air conditioning units. • Using top aisle containments reduces intensity of the hot spots by 5 °C. • Starting and ending racks' rows with cooling units prevents air by-passing. • Top and sides containments increases η r of terminal racks from 1.25 to 1.95. [ABSTRACT FROM AUTHOR]

• Blockchain-based decentralized workload management for Geo-distributed data centers. • Energy cost reduction of data centers seeing power outages and smart grid services. • A joint framework for minimal cost-based workload dispatching on data centers. • Comparative analysis of blockchain model with existing front-end server model. Minimizing the costs of cloud services is important for cloud service providers to grow their business. In cost calculation of cloud services, power consumption cost is a major factor. The variability present in renewable power generation, electricity price, and power demand variates the power utilization expenditures. Due to high computational requests, datacenters consume enormous energy every day, thereby necessitating the need for energy efficient designs. Therefore, an energy consumption cost reduction mechanism is required to optimize the data center's power consumption cost locally and to have the option to redistribute the workload on geo-distributed data centers in peak hours for further reduction in power consumption cost. In this paper, the authors investigate the energy consumption cost optimization problem in cloud datacenters and propose a blockchain-based decentralized workload distribution and management model. Moreover, we minimize request scheduler time for transferring any job from one data center to another and secure the energy cost optimization process due to shut down/ communication failure. The proposed work is the primitive to present the energy cost optimization problem in geo-distributed data centers by introducing blockchain-based secure workload scheduling method considering both the temporal and spatial disparities associated with electricity tariffs and workload arrival process. The simulations are evaluated on real-world workload of the Google data center and associated electricity tariffs. The blockchain-based decentralized workload management framework is evaluate based on privacy and security analysis, update overhead, latency, and throughput. Moreover, the results show blockchain model migrate workload in minimum 46 % reduced time compared to conventional model. [ABSTRACT FROM AUTHOR]

• A stochastic economic dispatch model in power systems with data centers is proposed. • The load of data centers is regulated to promote renewable integration. • Uncertainties of both renewables and DCs are considered. • Benders decomposition and a new type of group based multi-cut approach is applied to solve the problem. • The benefits of applying data centers in renewable integration is analyzed. The management of the demand side in the power system, such as data centers (DCs), can improve the stability of the grid and bring economic benefits in planning and operation. Considering the increasing applications of big data and net-based technologies, the computational jobs in different region can be shifted as flexible load. In this paper, a new type of decomposition method is proposed to solve the model of the stochastic economic dispatch (ED) incorporating DCs, which considers the uncertainties of job arriving rates in DCs and wind generation in power systems. The original ED formulations are decomposed with Benders decomposition to improve computational performance, and a new type of group based multi-cut Benders decomposition is applied to improve the convergence. Case studies on a joint 18-bus system illustrate the benefits of spacial coordination between DCs and power system. The comparison between single-cut and multi-cut approach proved the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Network virtualization is widely regarded as a key enabler to improve the desired flexibility and scalability of cloud computing services. With the rapid expansion of data centers, it remains a challenge to provide efficient management of resources for virtual machine placement. Due to multi-dimensional physical resources in servers, there maybe resource fragmentation causing by an imbalanced usage of multi-dimensional resources, when multiple virtual machines carrying diverse requests are deployed on the same server simultaneously. In this paper, we devote to how to balance multiple resources usage to alleviate resource fragmentation while maximizing the service rate for virtual machine placement, so that it can prevent waste of physical resources and undesirable performance. In order to solve such a bi-objective optimization problem, we present a joint bin-packing heuristic and genetic algorithm which achieves an approximate optimal solution with much lower time complexity. [ABSTRACT FROM AUTHOR]

• Co-optimization of server provisioning and power procurement. • Chance constraint is introduced to ensure quality of service guarantee. • Expected-value constraint is introduced to improve renewable energy utilization. • Dispatchable workloads can be deployed to increase renewable energy utilization. • Metrics are proposed to measure efficiency and sustainability of data centers. The reduction of greenhouse emissions is becoming a major goal of energy-intensive industries, such as data centers, and there have been significant efforts to achieve sustainable operations by meeting electricity consumption using renewable energy generations. Specifically, it has been a common practice for data centers to use renewable energy via on-site solar power generation to directly offset electricity consumption by renewable energy to contribute to environmental sustainability. However, the introduction of intermittent and non-dispatchable renewable energy generations for powering data centers that generally host time-varying workloads presents a significant challenge, and thus, this study mainly focuses on how to improve renewable energy utilization for data center operations considering the integration of co-located solar power generation and battery energy storage. The main objective is to develop a mathematical optimization model for energy-efficient and sustainable data center operations to minimize energy cost while ensuring the desired level of renewable energy utilization and the required quality of service guarantee. In particular, this study proposes a two-stage stochastic program integrated with an expected-value constraint and a chance constraint, and an integer programming and sampling-based approach are adopted to solve the problem to investigate optimal data center operations. The comprehensive numerical experiments are conducted to evaluate the proposed model compared with benchmark models for various parameter settings, and the results show that the proposed model can be successfully implemented to enable data centers to achieve the desired renewable energy utilization while improving energy efficiency. [ABSTRACT FROM AUTHOR]

• In this paper, we propose ThermoBench to evaluate the thermal efficiency of computing and storage clusters deployed in data centers. • We shed light on the criteria, metrics and challenges of developing a thermal efficiency benchmark. We pay particular attention to clusters running scalable client-server enterprise applications in data centers. • We characterize workload conditions in such a cluster computing environment in forms of client sessions of multiple transactional requests. To resemble real-world applications, ThermoBench makes use of the TPC-W benchmark to change transactional requests mix and load over time. • We apply ThermoBench to evaluate the thermal efficiency of a realworld cluster. The energy efficiency of a data center depends on the cooling cost of clusters in the data center. Enhancing thermal efficiency of clusters is a practical approach to reducing energy consumption cost, optimizing scalability, and improving reliability. In this paper, we propose ThermoBench to evaluate the thermal efficiency of computing and storage clusters deployed in data centers. We shed light on the criteria, metrics and challenges of developing a thermal efficiency benchmark. We pay particular attention to clusters running scalable client-server enterprise applications in data centers. Because these applications are quite popular in modern data centers, thermal efficiency benchmarks for clusters providing services to these applications become critical. We characterize workload conditions in such a cluster computing environment in forms of client sessions of multiple transactional requests. To resemble real-world applications, ThermoBench makes use of the TPC-W benchmark to change transactional requests mix and load over time. We apply ThermoBench to evaluate the thermal efficiency of a real-world cluster. Experimental results show that ThermalBench provides a simple yet powerful benchmark solution for assessing thermal behaviours of computing clusters in data centers as well as offers thermal-aware scheduling strategies during the course of requests dispatching. [ABSTRACT FROM AUTHOR]

This paper presents a statistical framework for predictive analysis of data center power usage effectiveness (PUE), with a focus on hyperscale data centers (HDCs). Thermodynamics-based PUE models considering representative economizer choices are proposed, taking climate variables and energy system parameters as inputs for robust PUE predictions. Sobol's method is used to assess total order sensitivity indices of key modeling parameters, suggesting that climate variables and uninterruptible power supply (UPS) efficiencies are the most important parameters. The PUE values of 17 HDCs operated by Google and Facebook were predicted, considering location-specific weather conditions, and uncertainties in energy system parameters and economizer choices. Results were verified using reported PUE values, indicating the model's effectiveness in capturing regional and seasonal PUE variations, and in generating point-estimations for macro-level data center (DC) energy models. Finally, achievable PUE values were computed through differential evolution, identifying minimum practical PUE values that could be obtained with state-of-the-art technologies. The framework can be applied in predictions of location-specific PUE values, PUE improvement analysis, and PUE target-setting by policy makers. • A statistical and thermodynamic model was developed to predict PUE values. • Climate parameters and UPS efficiency found to be the most important parameters. • PUE values of 17 Facebook and Google data centers were predicted and verified. • Practical minimum PUE values were identified by climate condition. • The model enables PUE estimation for use in regional/global energy system models. [ABSTRACT FROM AUTHOR]

This paper introduced and analyzed a new concept where an under-floor air supply (UFAD) system with cold aisle containment (CAC) is replaced by a new in-rack UFAD system called an in-rack cold aisle (IR-CA). The IR-CA system is analyzed using CFD simulation, and on-site measurement was carried out to validate the feasibility and reliability of simulation models. The study is divided into eight cases with seven different dimensions for the rack air inlet (2.2 m × 0.6 m, 0.2 m × 0.6 m, 0.3 m × 0.6 m, 0.4 m × 0.6 m, 0.5 m × 0.6 m, 0.6 m × 0.6 m, and 0.7 m × 0.6 m), while an additional partition plane is placed in Case 8 with a 0.6 m × 0.6-m in-rack air inlet. The thermal distribution is compared and analyzed in the eight cases, while cooling efficiency and energy saving is compared between the original and optimal cases. The results showed that the optimal thermal distribution is achieved in Case 8 with a 0.6 m × 0.6 m IR-CA and partition plane, while the thermal distribution in Case 8 with SAT of 23 °C is still much better than that in the original DC. The application of a 0.6 m × 0.6 m IR-CA and partition plane can save approximately 98 kW h/day in electricity consumption in the studied DC. A new evaluation index named the MS index is proposed to evaluate the optimization effects of the optimization model based on the original model. • A new concept of in-rack UFAD system and in-rack cold aisle (IR-CA) is proposed. • The thermal distribution can be improved when the width of IR-CA is larger than 0.3 m. • The optimal thermal distribution appears in the case of 0.6 m × 0.6 m IR-CA with partition plane. • The optimal model with SAT increased by 1 °C can still have better thermal distribution than the original model. • Approximate 98 kW h/day electricity can be saved in the optimal model. [ABSTRACT FROM AUTHOR]