Computing Data Centers Research Articles

Method for evaluation on energy consumption of cloud computing data center based on deep reinforcement learning

In view of the problem that the current method for evaluation on energy consumption of cloud computing data center has larger fluctuations of relative error and deviation value for evaluated power variation and greater evaluation training losses, this paper proposes an energy consumption evaluation method for cloud computing data center. Firstly, we build an energy consumption evaluation model, analyze the correlation between independent variables and dependent variables through multiple linear regression model and multiple nonlinear regression model, and then measure the accuracy of the evaluation results by using average relative deviation and relative deviation; The depth learning mechanism is introduced, and the depth residual network is used for iterative training of regression parameters to complete the mapping of regression parameters; Comprehensively obtain the regression parameters in the energy consumption model, put them into the memory playback pool, solve the energy consumption evaluation function according to the sub strategy parameters, and reconstruct the energy consumption model of cloud computing data center. The comprehensive evaluation of energy consumption of cloud computing data center is realized. The experimental results show that the energy consumption deviation of this method is small and stable. It can effectively reduce the relative error of evaluation power change and reduce the evaluation training loss.

Toward higher-radix switches with co-packaged optics for improved network locality in data center and HPC networks [Invited

We study the network locality improvements that can be achieved by using co-packaged optics in data center and high-performance computing (HPC) networks. The increased escape bandwidth offered by co-packaged optics can enable switches with speeds of 51.2 Tb/s and beyond. From a network architecture perspective, the key advantages of introducing co-packaged optics at the switch points include the implementation of large-scale topologies of > 12,000 end points with 4 × higher bisection bandwidth and the reduction of the required number of switches by > 40 % compared with state-of-the-art approaches. From a network operation perspective, improved network locality and faster operation can be achieved since the higher-radix switches can mitigate the impact of network contention. Placing applications under fewer leaf switches reduces the number of packets that cross the spine switches in a leaf-spine topology. The proposed scheme is evaluated via discrete-event simulations: we initially evaluate the network locality properties of the system by using virtual-machine traces from a production data center, and we subsequently quantify the performance improvements by simulating an all-to-all pattern for a variety of message sizes over a number of nodes. The results suggest that co-packaged optics form a promising solution for keeping up with bandwidth scaling in future networks. The virtual-machine analysis shows that large-scale applications can be placed under up to 50% fewer first-level switches, while the network analysis shows speedups of up to 7.1, which translates to execution time reductions of up to 26% and 42.7% for applications with communication ratios of 0.3 and 0.5, respectively.

Proactive Resilient Day-Ahead Unit Commitment With Cloud Computing Data Centers

To enhance the resilience of power systems toward the temporal and spatial impacts caused by extreme weather events, e.g., hurricanes, the flexibility of cloud data centers (CDCs) is treated as a kind of efficient demand response. Since the workloads of CDCs have the shifting capacity between different locations and time slots, a day-ahead unit commitment problem including data centers is proposed to explore the integrated spatial and temporal flexibility of the workloads to its full extent through task migration. Considering the uncertainty of probability density functions, the line failure rates, workload arrival rates, power loads are integrated into an ambiguity set. The scheduling process of generators and CDCs is modeled as a two-stage distributionally robust optimization problem, which is reformulated as a large-scale deterministic mixed-integer linear programming problem and solved by the multicuts Benders decomposition method. The performance of the proposed scheduling strategy is tested in both the IEEE 24-bus RTS system and the three-area RTS-96 system. The results reveal that the method could mitigate the adverse impacts of hurricanes by enhancing the resilience of power systems and decreasing the dropping workloads of CDCs.

Virtual Machine Consolidation using Roulette wheel selection Strategy

Cloud computing data centers are growing rapidly in both number and capacity to meet the increasing demands for highly-responsive computing and massive storage. Such data centers consume enormous amounts of electrical energy resulting in high operating costs and carbon dioxide emissions. The reason for this extremely high energy consumption is not just the quantity of computing resources and the power inefficiency of hardware, but rather lies in the inefficient usage of these resources. Virtual Machine [VM] consolidation involves live migration of VMs hence the capability of transferring a VM between physical servers with a close to zero down time. It is an effective way to improve the utilization of resources and increase energy efficiency in cloud data centers. VM consolidation consists of host overload/under load detection, VM selection and VM placement. In Our Proposed Model We are going to use Roulette-Wheel Selection Strategy, Where the VM selects the Instance type and Physical Machine [PM] using Roulette-Wheel Selection Mechanism Keywords—searchable encryption, dynamic update, cloud computing

A decision support model for waste heat recovery systems design in Data Center and High-Performance Computing clusters utilizing liquid cooling and Phase Change Materials

The Data Center (DC) and High-Performance Computing (HPC) sector is increasingly becoming one of the single largest energy consuming sectors in the energy system and the demand for new DCs is growing with the rising use of cloud-based services, social media and internet usage in general. DCs and HPC clusters employ energy intensive cooling in order to dissipate the generated heat from the Information Technology (IT) equipment. Conventionally, the cooling is almost exclusively done by air-based systems; however, the use of liquid based cooling has shown potential to increase computational efficiency of the systems and decrease cooling needs by operation above the free cooling limit. Furthermore, the liquid coolant can operate at higher temperatures allowing high temperature waste heat recovery. In addition, load shifting through Latent Thermal Energy Storage (LTES) will allow the non-controllable waste heat resource to be stored seasonally, thereby granting the ability to cover larger District Heating (DH) loads. In this work, a decision support model is developed that will take basic information regarding a HPC cluster or DC as inputs. The decision support model will provide a parameterized output that shows different configurations and design parameters that can be utilized for the system. The main outputs includes yearly energy savings, yearly cost savings and efficiency gains through the Power Usage Efficiency and the Energy Reuse Efficiency. The decision support model is demonstrated in a Danish case study. Electricity savings between 8.14 % and 10.8 % of the total cluster electricity consumption and a waste heat recovery potential of 85 MWh/year to 576 MWh/year are obtained. It is shown that if the DH covered is needed to be self-sufficient the configuration would require an LTES with PCM mass of 500 kg, but system configurations that operate as an addition to an existing local heating source shows increases in yearly energy savings of 332 % compared to self-sufficient configurations. The goal of the decision support model is to assist the design of future waste heat recovery applications through selection of system parameters including coolant temperatures, energy storage design parameters, DH supply temperatures and DH load coverage from the DC or HPC cluster.

O-Star: An Optical Switching Architecture Featuring Mode and Wavelength-Division Multiplexing for On-Chip Many-Core Systems

In this paper, we propose O-Star, a scalable optical switching architecture for on-chip many-core systems, employing hybrid mode and wavelength division multiplexing technology. The O-Star uses the Benes topology as the core switching module to realize non-blocking switching. Besides, we design a wavelength and mode allocation module to enable each processor to transmit data in parallel. To quantitatively analyze the minimum hardware cost required, we establish a mathematical model for the different number of processors. As a proof of concept, a 64-core optical switching architecture featuring 8-port Benes topology, 2-mode, and 4-wavelength channels is simulated with single channel 25 Gbps data rate. The O-Star is flexible to scale, both for the number of supported processors and the switching capacity. O-Star holds promise for realizing large-scale optical switching networks to address the incoming challenges of high-performance computing (HPC) systems and data center networks (DCNs).

Cone Model in Resource Provisioning for Service-Oriented Architecture System: An Effective Network Management to the Internet of Things

The rapid growth of technologies like the Internet of Things (IoT) has thrown newer challenges to resource management, making it one of the crucial tasks in dynamic application scenarios of IoT. Service-Oriented Architecture (SOA) plays an important role in this respect by proffering reference architecture towards better positioning of infrastructural software and hardware components, mostly comprehending from a service-oriented perspective. In this direction, we propose Cone Model that deals with the performance and handling issues with SOA system. Service integration is all important feature of SOA that supports communication and interaction between multiple service components (Service Containers). In addition to discussing the related work for critical resource allocation algorithms, this paper proposes Cone Model for assigning an available set of resources to requested services, explained through mathematical expressions and algorithms of the model. These algorithms justify that how different services from various devices in IoT can be provisioned on the adoption of this proposed cone model. It is illustrated through the estimated utilization of resources by the services. The proposed model aligns with the applications in multifarious domains, primarily in the IoT and in areas of resource sharing for cloud computing and massive data centers.

Power- and QoS-Aware Job Assignment With Dynamic Speed Scaling for Cloud Data Center Computing

As the usage of mission-critical mobile applications increases in Industry 4.0, such as smart manufacturing and self-driving cars, the cloud computing paradigm and its supporting data centers have become more crucial. However, a common practice in the cloud data center computing industry tends to supply a surfeit of computing resources mainly for a robust quality-of-service (QoS). In this paper, we propose a simple real-time algorithm which combines a power-aware job assignment policy for a centralized job dispatcher and a power- and QoS-aware dynamic speed scaling policy for each physical machine (PM). The job assignment policy is called “Join the Least Power Consuming (LPC) Server” that routes an incoming cloud job to a server spending minimum power upon request. The server-side adaptive speed scaling policy expedites energy efficiency and satisfies response time-associated QoS condition. We call this policy “Minimizing Earliness (ME)” since it manages the server speed towards finishing jobs at their deadlines as precisely as possible, reducing the earliness of job completion. The design principle of LPC-ME combination supports both energy efficiency and service quality required in cloud data centers. Numerical experiments compare the proposed algorithm’s power consumption and response time with those of existing popular policies and demonstrate better energy efficiency with negligible degradation of service quality.

Features of the infocommunication infrastructure development management mecha-nism

In the context of the Industrial Revolution 4.0 and the total digitalization of socio-economic processes in society, research related to the development and justification of effective mechanisms for managing the innovative development of modern infocommunication infrastructure and ensuring a high level of efficiency of the practical application of infocommunication technologies (ICT) is relevant. The most important feature of the evolution of such infrastructure and the ICTs themselves is their significant correlation with the totality of processes carried out in the global and national economies. In fact, ICTs play the role of a catalyst for the above-mentioned global and local processes, predetermine the pace of scientific and technological progress. An analysis of the achievements of a number of economically developed countries over the past few decades indicates that the state of a single national economy and the dynamics of its processes directly depend on the pace of introduction of modern means of communication and telecommunications, computer networks, data centers and other innovations into all spheres.
 In connection with the above, a number of global problems and local tasks arise regarding not only the adequate determination of the necessary level of development of the infocommunication infrastructure, due to the formation of new socio-economic realities, and forecasting the likely consequences of intensive use of ICT tools in various fields of activity, but also the formation of mechanisms for managing innovative development. These mechanisms should be effective and efficient in achieving the relevant management objectives.
 The article pays attention to a number of aspects and features of the mechanism proposed by the authors for managing the development of the infocommunication infrastructure for the formation of the digital economy and the information society as a whole. This mechanism is based on the practical application of specific methodological tools that allow monitoring the status and potential of the infocommunication infrastructure through a comprehensive assessment of the effectiveness of the introduction of ICT. The mechanism developed by the authors for managing the development of the infocommunication infrastructure will allow public authorities to specify the strategy and tactics of forming the foundations of the digital economy and the information society as a whole.

Research on information classification and storage in cloud computing data center based on group collaboration intelligent clustering

In order to overcome the problems of long time and low accuracy of traditional methods, a cloud computing data center information classification and storage method based on group collaborative intelligent clustering was proposed. The cloud computing data center information is collected in real time through the information acquisition terminal, and the collected information is transmitted. The optimization function of information classification storage location was constructed by using the group collaborative intelligent clustering algorithm, and the optimal solutions of all storage locations were evolved to obtain the elite set. According to the information attribute characteristics, different information was allocated to different elite sets to realize the classified storage of information in the cloud computing data center. The experimental results show that the longest time of information classification storage is only 0.6 s, the highest information loss rate is 10.0%, and the highest accuracy rate is more than 80%.

Energy efficiency in cloud computing data center: a survey on hardware technologies

Energy efficiency in cloud computing data center: a survey on hardware technologies

CPU Overclocking: A Performance Assessment of Air, Cold Plates, and Two-Phase Immersion Cooling

Computing capacity has always been on the upward climb due to the constant technological improvements in semiconductor manufacturing and packaging industry. This growth in computing capability is usually accompanied by a steep rise in heat flux density associated with the electronic component (CPUs or GPUs, for example). High-performance computing (HPC) data centers often employ several of these high-performance devices for crunching their enormous artificial intelligence (AI) or scientific computing workloads. State-of-the-art air cooling technologies throttle after a certain heat flux level and would require bigger heat sinks driving enormous airflow through it, which may not be desirable from a data center operation standpoint. Hence, a lookout for advanced thermal management techniques is quite imperative. In this article, a commercially available intel overclockable CPU i9-9900k was exercised with high performance workloads while employing and evaluating three different cooling technologies namely air-cooled, cold plates, and two-phase (2P) immersion. The high heat carrying capacity associated with cold plates and 2P immersion techniques outperformed the air-cooled solution by constantly yielding higher CPU clock rates up to 41% utilizing cold plates and 51% employing 2P immersion. The performance of cold plates and 2P immersion was evaluated at different functional points (different coolant operating temperatures and 2P coolant types for example) to understand how the technologies would respond in a legacy or modern data center operation. It was again observed that both 2P immersion technology and cold plates provided the least thermal resistance (as low as 0.247 °C/W) path to heat transfer and therefore provided higher computational performance and efficiency compared to air cooling. Further improvement in performance is not limited by cooling but by packaging constraints. Advanced packaging techniques coupled with minimizing the interfacial resistance can reap enhanced performance and energy efficiency gains using these modes of cooling. This article demonstrates the potential increase in virtual machine (VM) performance that can be attained using cold plates or 2P immersion. Authors recommend adoption of liquid cooling in early chip design and architecture phases. Finally, the article provides a generic overclocking methodology that the industry can use to qualify postturbo server class chip performance.

Energy Efficient and VM Consolidation Framework using Improved Spider Monkey Optimization Algorithm

Cloud Computing is rapidly being utilized to operate informational technological services by outstanding technologies for a variety of benefits, including dynamically improved resources planning and a new service delivery method. The Cloud computing process is occurred by allowing the client devices for data access through the internet from a remote server, computers, and the databases. An internet connection is linked among the front end users such as client device, network, browser, and software application with the back end that constitutes of servers, computers, and database. For satisfying the demands of the Service Level Agreement (SLA), providers of cloud service should reduce the usage of energy. Capacity reservations oriented system is available by clouds’ providers to permit users for customizing Virtual Machines (VMs) having specified age and geographic resources, reduces the amount to be paid for cloud services. To overcome the aforementioned issue, an Improved Spider Monkey Optimization (ISMO) approach is proposed for cloud center optimization. The VM consolidation architecture based on the proposed ISMO algorithm decreases energy usage while attempting to prevent Service Level Agreement breaches. The accessibility of hosts or virtual machines (VMs) for task performance is measured by fitness. If the number of tasks to be handled increases the hosts of VMs available at right state. The proposed VM consolidation architecture decreases energy usage while also attempting to prevent Service Level Agreement breaches and also provide energy-efficient computing in data centers. The proposed approach may be utilized to provide energy-efficient computing in data centers. The energy efficiency of the proposed ISMO method is achieved 28266 whereas, the existing algorithm showed an energy efficiency of 6009 and 10001.

Research on Energy Saving of Air Conditioning System in Data Center

With the development of the Internet today, there are more and more small and medium data center computer rooms, the energy consumption of data centers has also become the focus of attention. Air conditioning system is a big energy consumer in data centers, approximately 40% of the total energy consumption of the data center. So reducing the energy consumption of air conditioners has become the top priority of reducing the energy consumption of data centers. This paper studies the air distribution generated by different air-conditioning methods, building a computer room model and uses the simulation software 6sigma to simulate the different air conditioning modes, thereby reducing the energy consumption of the computer room.

Globally Scheduling Volunteer Computing

Volunteer computing uses millions of consumer computing devices (desktop and laptop computers, tablets, phones, appliances, and cars) to do high-throughput scientific computing. It can provide Exa-scale capacity, and it is a scalable and sustainable alternative to data-center computing. Currently, about 30 science projects use volunteer computing in areas ranging from biomedicine to cosmology. Each project has application programs with particular hardware and software requirements (memory, GPUs, VM support, and so on). Each volunteered device has specific hardware and software capabilities, and each device owner has preferences for which science areas they want to support. This leads to a scheduling problem: how to dynamically assign devices to projects in a way that satisfies various constraints and that balances various goals. We describe the scheduling policy used in Science United, a global manager for volunteer computing.

Feature Matching Synchronized Reasoning from Energy-Based Memory Network for Intelligent Data Management in Cloud Computing Data Center

A cloud data center for software-as-a-service (SaaS) was built for the purpose of stably managing these server computers in one place in order to provide an uninterrupted service, not only for a stable power supply and security but also for the efficient data management. To manage such a data center efficiently, it is important to build a cloud database with structured storage above all else. In recent decades, many studies have focused on designing cloud data centers and most of the research has focused on communication traffic, routing, topological issues and communication technology. However, in order to build an efficient cloud database that can support user demand, the most sophisticated intelligent system, based on AI technology and considering user convenience, should be designed. From this viewpoint, adopting human brain functions, Energy-Based Memory Network was designed for a knowledge-based frame of an intelligent system. And its event-related synchronized data extraction mechanism was proposed. In particular, a Thinking Thread extraction phase was implemented for the reasoning process using qualia matching and a deep extraction method in a cloud database. The purpose of this approach is to design and implement an intelligent cloud database that has an efficient structure and mechanism for supporting user demand and providing accurate, prompt services. In experiments, the working phase of the functions was simulated with data and analyzed. As a result, it was confirmed that the proposed system works well and intelligently for the design purpose.

Corrosion Study on Single-Phase Liquid Cooling Cold Plates With Inhibited Propylene Glycol/Water Coolant for Data Centers

Abstract Single-phase cold plate liquid cooling attracts more and more attention to high-performance computing (HPC), cloud computing, and general computing data centers for the thermal management of modern microprocessors and adjacent components due to liquid’s inherent advantage of higher specific heat compared to air. Deionized (DI) water is usually used as a coolant for liquid cooling in data centers. On the contrary, propylene glycol/water is recommended as a coolant for single-phase cold plate liquid cooling in this study for following reasons. The inhibited propylene glycol-based fluids of 25+% vol have the benefit of being biostatic and not requiring addition of biocides. They also offer freezing protection in the usage of data centers in cold climates. The cold plates made from copper is prone to oxide even under the room temperature, and the dissimilarity between brazing material and copper can also cause galvanic corrosion in the usage. In this paper, a study was carried out to investigate cold plate corrosion with inhibited propylene glycol/water using design of experiments (DOE) method. This study shows manufacturing process plays an important role on corrosion of cold plates made from copper and the corrosion risk can be mitigated by enabling new manufacturing processes, including friction stir welding (FSW) and nickel plating to the inside surface of the cold plate in the manufacturing process.

Performance Analysis of Packet Aggregation Mechanisms and Their Applications in Access (e.g., IoT, 4G/5G), Core, and Data Centre Networks.

The transmission of massive amounts of small packets generated by access networks through high-speed Internet core networks to other access networks or cloud computing data centres has introduced several challenges such as poor throughput, underutilisation of network resources, and higher energy consumption. Therefore, it is essential to develop strategies to deal with these challenges. One of them is to aggregate smaller packets into a larger payload packet, and these groups of aggregated packets will share the same header, hence increasing throughput, improved resource utilisation, and reduction in energy consumption. This paper presents a review of packet aggregation applications in access networks (e.g., IoT and 4G/5G mobile networks), optical core networks, and cloud computing data centre networks. Then we propose new analytical models based on diffusion approximation for the evaluation of the performance of packet aggregation mechanisms. We demonstrate the use of measured traffic from real networks to evaluate the performance of packet aggregation mechanisms analytically. The use of diffusion approximation allows us to consider time-dependent queueing models with general interarrival and service time distributions. Therefore these models are more general than others presented till now.

Moisture Computing-Based Internet of Vehicles (IoV) Architecture for Smart Cities.

Recently, the concept of combining ‘things’ on the Internet to provide various services has gained tremendous momentum. Such a concept has also impacted the automotive industry, giving rise to the Internet of Vehicles (IoV). IoV enables Internet connectivity and communication between smart vehicles and other devices on the network. Shifting the computing towards the edge of the network reduces communication delays and provides various services instantly. However, both distributed (i.e., edge computing) and central computing (i.e., cloud computing) architectures suffer from several inherent issues, such as high latency, high infrastructure cost, and performance degradation. We propose a novel concept of computation, which we call moisture computing (MC) to be deployed slightly away from the edge of the network but below the cloud infrastructure. The MC-based IoV architecture can be used to assist smart vehicles in collaborating to solve traffic monitoring, road safety, and management issues. Moreover, the MC can be used to dispatch emergency and roadside assistance in case of incidents and accidents. In contrast to the cloud which covers a broader area, the MC provides smart vehicles with critical information with fewer delays. We argue that the MC can help reduce infrastructure costs efficiently since it requires a medium-scale data center with moderate resources to cover a wider area compared to small-scale data centers in edge computing and large-scale data centers in cloud computing. We performed mathematical analyses to demonstrate that the MC reduces network delays and enhances the response time in contrast to the edge and cloud infrastructure. Moreover, we present a simulation-based implementation to evaluate the computational performance of the MC. Our simulation results show that the total processing time (computation delay and communication delay) is optimized, and delays are minimized in the MC as apposed to the traditional approaches.

Assessing anthropogenic heat flux of public cloud data centers: current and future trends.

Global average temperature had been significantly increasing during the past century, mainly due to the growing rates of greenhouse gas (GHG) emissions, leading to a global warming problem. Many research works indicated other causes of this problem, such as the anthropogenic heat flux (AHF). Cloud computing (CC) data centers (DCs), for example, perform massive computational tasks for end users, leading to emit huge amounts of waste heat towards the surrounding (local) atmosphere in the form of AHF. Out of the total power consumption of a public cloud DC, nearly 10% is wasted in the form of heat. In this paper, we quantitatively and qualitatively analyze the current state of AHF emissions of the top three cloud service providers (i.e., Google, Azure and Amazon) according to their average energy consumption and the global distribution of their DCs. In this study, we found that Microsoft Azure DCs emit the highest amounts of AHF, followed by Amazon and Google, respectively. We also found that Europe is the most negatively affected by AHF of public DCs, due to its small area relative to other continents and the large number of cloud DCs within. Accordingly, we present mean estimations of continental AHF density per square meter. Following our results, we found that the top three clouds (with waste heat at a rate of 1,720.512 MW) contribute an average of more than 2.8% out of averaged continental AHF emissions. Using this percentage, we provide future trends estimations of AHF densities in the period [2020–2100]. In one of the presented scenarios, our estimations predict that by 2100, AHF of public clouds DCs will reach 0.01 Wm−2.