Conventional Data Center Research Articles

Evaluation of a solar-powered absorption cooling system to a data center

Nowadays, data processing is a fundamental operation for modern businesses such as banks, technology companies, and factories, among others. However, computers dissipate significant amounts of heat yielding to an operational temperature rise. Considering that these machines cannot operate properly in inappropriate temperatures or at extreme conditions, they can come to a stop due to overheating. Consequently, cooling and air-conditioning systems are necessary to keep the proper operating temperature as well as the room temperature itself. On the other hand, a data center air-conditioning system drains a large amount of electrical power. Based on this, this paper evaluates a solar-powered absorption cooling system to assist the traditional electric chiller system resulting in energy saving, an advantage over conventional cooling, and day availability for this system. A case study is analyzed in a conventional data center located in the city of Sao Paulo, Brazil. First, the electric power density consumed by computers is 2.0 kW/m2, which represents a typical power load of an IT room. In addition, some other power density cases are also analyzed, namely: 0.5, 1.0, 4.0, and 8.0 kW/m2; these would occur at partial or at high operational loads. Local solar irradiation indexes are based on ASHRAE temperature and solar data for that city. The results are valid for a typical year and are compared to (1) a conventional data center and (2) the event combined with the solar-powered cooling system.

Temperature distribution estimation via data-driven model and adaptive Kalman filter in modular data centers

With the rapid development of information and communications technology, increasing number of data centers is required to support the cloud computing, and critical web-based services that run our daily lives. The conventional cloud data centers usually adopt computer room air conditioner or inRow units as the cooling sytem, while the rack mountable cooling unit is a more promising equipment due to the economy, exact controllability, flexibility, and scalability. To ensure the efficiency of control system in rack mountable cooling unit and the security of servers in the data centers, the information of temperature distribution is very essential. Basically, the temperature distribution could be obtained through physical sensors easily. However, considering the cost of whole system and the burden of fault diagnosis in sensor networks, the number of temperature sensors should be kept down to a bare minimum. Therefore, it is necessary to develop an effective and real-time observer to estimate the temperature distribution in the system. Besides, due to the complex air flow and heat transfer in the container, it is quite difficult to construct a physics model. To this end, a novel observer embracing data-driven model and adaptive Kalman filter is proposed in this work. Auto regression exogenous model is adopted as the framework of data-driven model, and the model is identified through a algorithm of partial least square. Moreover, to represent the nonlinear behaviors in the system, fuzzy c-means is applied for data classification and getting multiple local linear models. Finally, adaptive Kalman filter is utilized to estimate the temperature distribution on the basis of proposed data-driven model. The estimation results based on experimental data indicate the performance of proposed approach is remarkable.

Energy Efficient Distributed Processing for IoT

In the near future, the number of Internet connected objects is expected to be between 26 - 50 billion devices. This figure is expected to grow even further due to the production of miniaturized portable devices that are lightweight, energy, and cost-efficient. In this article, the entire IoT-fog-cloud architecture is modeled, the service placement problem is formulated using Mixed Integer Linear Programming (MILP) and the total power consumption is jointly minimized for processing and networking. We evaluate the distributed processing paradigm for both the un-capacitated and capacitated design settings in order to provide solutions for the long-term and short-term basis, respectively. Furthermore, four aspects of the IoT processing placement problem are examined: 1) IoT services with non-splittable tasks, 2) IoT services with splittable tasks, 3) impact of processing overheads needed for inter-service communication and 4) deployment of special-purpose data centers (SP-DCs) as opposed to the conventional general-purpose data center (GP-DC) in the core network. The results showed that for the capacitated problem, task splitting introduces power savings of up to 86% compared to 46% with non-splittable tasks. Moreover, it is observed that the overheads due to inter-service communication greatly impacts the total number of splits. However much insignificant the overhead factor, the results showed that this is not a trivial matter and hence much attention needs to be paid to this area to make the best use of the resources that are available at the edge of the network.

A new strategy based on approximate dynamic programming to maximize the net revenue of IaaS cloud providers with limited resources

Abstract With rapid development of cloud computing, more conventional Internet Data Centers get involved into this business model. Providers of small and medium sized data centers have relatively limited computing resources compared to global IaaS providers such as Amazon. The demand for computing at these service providers may be above the available capacity especially during peak shopping periods (e.g. mother’s day or singles’ day). However, it is undesirable to blindly expand capacity, which will not only incur the purchase cost of new machines but also the maintenance costs associated with resource over-provisioning. For these service providers, the most important thing is how to effectively manage resources during peak demand periods to maximize their net revenue. In this context, we focus on the joint admission control and virtual machine placement problem, and propose a dynamic optimization model considering that customers dynamically create and terminate virtual machines in cloud computing environment. In terms of problem complexity, the developed dynamic optimization model typically suffers from “curse of dimensionality” and is computationally intractable. So we resort to approximate dynamic programming framework and propose a new strategy to yield a tractable model for real-time decision-making. Extensive simulations are conducted to compare our proposed strategy with other strategies including existing methods of virtual machine placement. The simulation results demonstrate that our customized strategy achieves substantial revenue improvements and the improvement is more significant as resource provisioning is tighter. In addition, it can be scalable to solve large cases up to 1000 physical machines and yield a real-time decision in a reasonable time.

Influence of cooling architecture on data center power consumption

Almost thirty percent of the power consumed by data centers (DCs) is attributable to the cooling of IT equipment (ITE). There are opportunities to reduce a DC's energy budget by considering alternatives to traditional cooling methods, which experience inherent airflow deficiencies due to hot air recirculation and cold air bypass. Minimizing these two air distribution problems results in more effective cooling, but the two effects are manifest differently in the three conventional DC cooling architectures, i.e., (a) room-based, (b) row-based, and (c) rack-based cooling. Despite the intuitive logic that predicts improved cooling air distribution within row- and rack-based architectures that include shorter airflow pathlengths compared to room-based systems that have longer paths, the mechanism through which improvements translate into energy savings is not well understood. Therefore, we present methodologies that resolve the characteristic airflow and temperature distributions for three cooling architectures using computational fluid dynamics. These results inform thermodynamics models of the power consumptions that are required to cool these three architectures. The analysis reveals that row- and rack-based architectures reduce cooling power by much as 29% over a room-based architecture. Adding an enclosure within row- and rack-based architectures to separate the hot and cold airflows provides further 18% reduction in cooling power. This analysis facilitates better DC design from a cooling power consumption perspective.

Equation-based object-oriented modeling and simulation of data center cooling systems

In this paper, we introduce a newly developed open source data center package in the Modelica Buildings library to support modeling and simulation of cooling and control systems of data centers. The data center package contains major thermal and control component models, such as Computer Room Air Handler, Computer Room Air Conditioner, models of different subsystem configurations such as chillers with differently configured waterside economizers, as well as templates for different systems. Two case studies based on this package are performed to investigate the performances of the cooling and electrical system under normal conditions and emergency situations such as a blackout: one is for a data center powered by conventional energy, and the other is for a data center powered by renewable energy. Simulation results show that the dynamic modeling and multi-domain simulation in the Modelica-based tool make it convenient for users to investigate both normal and emergent operations for conventional and renewable data centers.

A methodology to assess the availability of next-generation data centers

Cloud data center providers benefit from software-defined infrastructure once it promotes flexibility, automation, and scalability. The new paradigm of software-defined infrastructure helps facing current management challenges of a large-scale infrastructure, and guarantying service level agreements with established availability levels. Assessing the availability of a data center remains a complex task as it requires gathering information of a complex infrastructure and generating accurate models to estimate its availability. This paper covers this gap by proposing a methodology to automatically acquire data center hardware configuration to assess, through models, its availability. The proposed methodology leverages the emerging standardized Redfish API and relevant modeling frameworks. Through such approach, we analyzed the availability benefits of migrating from a conventional data center infrastructure (named Performance Optimization Data center (POD) with redundant servers) to a next-generation virtual Performance Optimized Data center (named virtual POD (vPOD) composed of a pool of disaggregated hardware resources). Results show that vPOD improves availability compared to conventional data center configurations.

Flexible fine-grained baseband processing with network functions virtualization: Benefits and impacts

The increasing demand for wireless broadband connectivity is leading mobile network operators towards new means to expand their infrastructures efficiently and without increasing the cost of operation. Network Functions Virtualization (NFV) is a step towards virtualization-based, low-cost flexible and adaptable networking services. In the context of centralized baseband architectures, virtualization is already employed to run baseband processing units as software on top of conventional data center hardware. However, current virtualization solutions consider atomic virtualization, i.e., single virtual machines implementing all baseband functionalities. In this article, we propose the fine-grained virtualization of baseband processing to achieve a more flexible distribution of the processing workload in centralized architectures. We also evaluate the benefits of our approach in terms of: (i) The bandwidth requirements for each fine-grained distribution option, (ii) the latency experienced by mobile users for each fine-grained distribution option, and (iii) the total CPU usage of each fine-grained baseband processing function.

Load Balance in Data Center SDN Networks

<span>In the last two decades, networks had been changed according to the rapid changing in its requirements. The current Data Center Networks have large number of hosts (tens or thousands) with special needs of bandwidth as the cloud network and the multimedia content computing is increased. The conventional Data Center Networks (DCNs) are highlighted by the increased number of users and bandwidth requirements which in turn have many implementation limitations. The current networking devices with its control and forwarding planes coupling result in network architectures are not suitable for dynamic computing and storage needs. Software Defined networking (SDN) is introduced to change this notion of traditional networks by decoupling control and forwarding planes. So, due to the rapid increase in the number of applications, websites, storage space, and some of the network resources are being underutilized due to static routing mechanisms. To overcome these limitations, a Software Defined Network based Openflow Data Center network architecture is used to obtain better performance parameters and implementing traffic load balancing function. The load balancing distributes the traffic requests over the connected servers, to diminish network congestions, and reduce underutilization problem of servers. As a result, SDN is developed to afford more effective configuration, enhanced performance, and more flexibility to deal with huge network designs</span>

A cost-effective low-latency overlaid torus-based data center network architecture

In this paper, we present the design, analysis, and implementation of a novel data center network architecture named CLOT, which delivers significant reduction in the network diameter, network latency, and infrastructure cost. CLOT is built based on a switchless torus topology by adding only a number of most beneficial low-end switches in a proper way. Forming the servers in close proximity of each other in torus topology well implements the network locality. The extra layer of switches largely shortens the average routing path length of torus network, which increases the communication efficiency. We show that CLOT can achieve lower latency, smaller routing path length, higher bisection bandwidth and throughput, and better fault tolerance compared to both conventional hierarchical data center networks as well as the recently proposed CamCube network. Coupled with the coordinate based geographical addresses and credit based flow control, the specially designed POW routing algorithm helps CLOT achieve its maximum theoretical performance. Besides, an automatic address configuration mechanism and malfunction detection mechanism are provided to facilitate the network deployment and configuration. The sufficient mathematical analysis and theoretical derivation prove both guaranteed and ideal performance of CLOT.

In Compute/Memory Dynamic Packet/Circuit Switch Placement for Optically Disaggregated Data Centers

Network function services on conventional hybrid data center (DC) architectures such as HELIOS arehard-wiredanddedicatedto specificnetworkresources. This limits flexibility and performance to handle diverse traffic patterns. Furthermore, disaggregation of server resources has shown promising potential to improve resource utilization, which has been a limitation of conventional server-centric DCs. This paper presents a reconfigurable hybrid disaggregated DC (dRedBox) architecture that combines the concept of server resource disaggregation with cutting-edge software and electronic and optical technologies. ThedRedBoxarchitecture provides a remarkable amount of flexibility and connectivity through hardware-based multilayer network function service programmability. This allows for multilayer network services to be dynamically deployed at runtime to network resources and, in turn, handle diverse traffic patterns. Furthermore, this study proposes algorithms and strategies for selecting and deploying electronic packet switching and optical circuit switching function services to implement virtual machine network requests across dRedBox and conventional hybrid disaggregated architectures under different traffic patterns. Finally, the performance of the various strategies on the dRedBox and conventional hybrid disaggregated DC architectures is evaluated in terms of blocking probability, energy efficiency, network utilization, and cost. Extensive results show that, at 10% blocking probability, dRedBox architecture achieves 100% gain on VM placement and 92% energy savings compared with conventional hybrid disaggregated architectures.

Online-Batch Joint Routing and Channel Allocation for Hybrid Data Center Networks

To cope with the unprecedented traffic explosion, Internet giants are urged to rethink their data center design. Unfortunately, the conventional wired data centers struggle to support the impressive growth of both data and online services. In this regard, we resort to augmenting the wired data center network (DCN) with wireless communication (60 GHz technology). Heretofore, only few research work have dealt with the optimization of multi-hop communications in such hybrid DCN (HDCN) infrastructures. In this paper, we investigate, in HDCN, the issue of jointly: 1) routing and 2) channel allocating, in both online and batch arrival modes. First, we formulate the online joint routing and channel assignment problem as a minimum weighted perfect matching problem. We propose a new strategy named joint routing and channel allocation algorithm that makes use of the Blossom algorithm and sequentially computes the optimal routing communication paths. Second, to handle the batched arrivals of flows, we formulate the batch joint routing and channel assignment problem as an advanced multi-commodity flow model. We propose two scalable approaches: 1) a heuristic based solution, named joint routing and channel assignment heuristic and 2) an approximate solution, named scalable joint batch routing and channel allocation, based on the Lagrangian relaxation technique. Based on extensive network simulations conducted in QualNet simulator while considering the full protocol stack, the obtained results for both: 1) real Facebook’s data center and 2) uniform, traces, show that our schemes outperform the related strategies.

Mathematical Optimization Modelling for Fast-Switched and Delay Minimized Scheduling for Intra-Cell Communication in an AWGR-Based PON Data Center

As the internet traffic along with the processing power in data centers are exponentially growing, the need for the design of energy efficient with highly elastic networking infrastructure to support the different applications and cloud services that can be hosted in data centers have become a hot research area. A key departure from the norm is that conventional routers and switches in conventional data centers are replaced with high performance Passive Optical Networks (PONs) to take over the role of routing and traffic forwarding through efficient resource provisioning algorithms. In this paper, the different aspects of PONs in the design of energy efficient, high capacity, and highly elastic networking infrastructures to support the applications and services hosted by modern data centers are considered. In this work, a mathematical optimization model for energy-efficient and delay-minimized scheduling in AWGR based PON data center for PON cell fabric configuration will be presented. The performance of the proposed architecture in terms of efficient scheduling against average delay and power consumption for different traffic loads and patterns will be evaluated. Different scenarios of traffic; random and unbalanced with hotspots are examined to evaluate the average delay and power consumption with and without sleep mode. Results have shown that with sleep mode enabled, power savings for two evaluated objective functions have shown similar results when examining different traffic patterns. The power savings range between 8% and 55% during low and high load activities, respectively. However, minimization of delay model has shown improvement in reducing total average delay reaching up to 42% if compared with the model with objective of minimization of power consumption.

Green Cloud and Virtual Machines Migration Challenges

Cloud computing, currently, is one of the most efficient technologies over the IT and global application environments. This new environments has become one the major an effective working environment for small and medium scale enterprises. No one can deny the importance of the cloud computing, but replacing conventional data centers by cloud data centers need more attention. One of the important sides of cloud data centers is down time. To achieve this approach, data centers need to be active all the time and this approach needs more power. Thus, one of our concerns is about consuming more energy, because more energy needs more energy production. Therefore, traditional cloud data centers need huge electricity energy for running servers and cooling systems for example. Electricity production need carbon-based fuels, principally coal, oil, and natural gas cause the emission of the greenhouse gases like carbon dioxide (CO2). This emission may cause the global warming that has direct effects on humankind environment. To decrease greenhouse gases, it needed to consume less carbon-based fuels and green cloud data centers based on less consuming energy that known as high-efficient data centers. In green cloud we try to reduce the number of active devices and consume less electricity energy. The green data centers toke an advantage of the VMs ability of copying, deleting and moving across the data center. This action over servers may cause serious effect over all part of data centers in designing green data centers and in this paper bolds these challenges on this path.

Reliability and Survivability Analysis of Data Center Network Topologies

The architecture of several data centers have been proposed as alternatives to the conventional three-layer one.Most of them employ commodity equipment for cost reduction. Thus, robustness to failures becomes even more important, because commodity equipment is more failure-prone. Each architecture has a different network topology design with a specific level of redundancy. In this work, we aim at analyzing the benefits of different data center topologies taking the reliability and survivability requirements into account. We consider the topologies of three alternative data center architecture: Fat-tree, BCube, and DCell. Also, we compare these topologies with a conventional three-layer data center topology. Our analysis is independent of specific equipment, traffic patterns, or network protocols, for the sake of generality. We derive closed-form formulas for the Mean Time To Failure of each topology. The results allow us to indicate the best topology for each failure scenario. In particular, we conclude that BCube is more robust to link failures than the other topologies, whereas DCell has the most robust topology when considering switch failures. Additionally, we show that all considered alternative topologies outperform a three-layer topology for both types of failures. We also determine to which extent the robustness of BCube and DCell is influenced by the number of network interfaces per server.

Soft Failures in Large Datacenters

A major problem in managing large-scale datacenters is diagnosing and fixing machine failures. Most large datacenter deployments have a management infrastructure that can help diagnose failure causes, and manage assets that were fixed as part of the repair process. Previous studies identify only actual hardware replacements to calculate Annualized Failure Rate (AFR) and component reliability. In this paper, we show that service availability is significantly affected by soft failures and that this class of failures is becoming an important issue at large datacenters with minimum human intervention. Soft failures in the datacenter do not require actual hardware replacements, but still result in service downtime, and are equally important because they disrupt normal service operation. We show failure trends observed in a large datacenter deployment of commodity servers and motivate the need to modify conventional datacenter designs to help reduce soft failures and increase service availability.

Cooperative control architecture of fan-less servers and fresh-air cooling in container servers for low power operation

In order to minimize the container server power consumption, a new cooling system that incorporates fan-less servers and freshair cooling is proposed. In a conventional container data center, the required air flow for sever cooling is supplied by both server built-in fans and container facility fans. Therefore, this work has been carried out on fan-less servers to reduce power consumption. Although fanless servers are expected to reduce power consumption, facility fans have to provide excessive air to secure a safe operation of servers. In order to achieve optimized air-flow from facility fans to cool fan-less servers, a power saving control system incorporating the IT system and cooling facilities is proposed. Here, facility fans are controlled based on server information such as CPU temperature, rack position and so on. Through this study, we suggest that the minimum point in total power consumption of the container server with no performance penalty existed by the trade-off relationship between the power consumption changes of servers and of facility fans with CPU temperature. This enables us to operate the server system with minimized power consumption depending on the air temperature. To verify the energy-saving effect of this technology, a prototype container server with the proposed system was constructed. As a result, 22.8% energy saving was achieved with this new system, compared with the conventional container servers with built-in fans.

On the Feasibility of Completely Wirelesss Datacenters

Conventional datacenters, based on wired networks, entail high wiring costs, suffer from performance bottlenecks, and have low resilience to network failures. In this paper, we investigate a radically new methodology for building wire-free datacenters based on emerging 60-GHz radio frequency (RF) technology. We propose a novel rack design and a resulting network topology inspired by Cayley graphs that provide a dense interconnect. Our exploration of the resulting design space shows that wireless datacenters built with this methodology can potentially attain higher aggregate bandwidth, lower latency, and substantially higher fault tolerance than a conventional wired datacenter while improving ease of construction and maintenance.

The case for a rigorous approach to automating software operations and management of large-scale sensor networks

Software Operations and Management (O&M) i.e., installing, configuring, and updating thousands of software components within a conventional Data Center is a well-understood issue. Existing frameworks such as the Rocks toolkit have revolutionized the way system administrators deploy and manage large-scale compute clusters, storage servers, and visualization facilities. However, existing tools like Rocks are designed for a "friendly" Data Center environment where stable power along with high-performance compute, storage, and networking is the norm. In contrast, sensor networks are embedded deeply within the harsh physical environment where node failures, node mobility and idiosyncrasies of wireless networks are the norm. In addition, device heterogeneity and resource-constrained nature (e.g., power, memory, CPU capability) of the sensor cyberinfrastructure (CI) are realities that must be addressed and reconciled. Although sensor CI must be more adaptable and more-rapidly reconfigurable than the data center equivalents, few if any of the existing software O&M tools and techniques have been adapted to the significantly more challenging environment of sensor networks. A more automated approach to software O&M would provide significant benefits to system builders, operators, and sensor network researchers. We argue that by starting with software O&M techniques developed for data centers, and then adapting and extending them to the world of resource-constrained sensor networks, we will be able to provide robust and scientifically reproducible mechanisms for defining the software footprint of individual sensors and networks of sensors. This paper describes the current golden-image based software O&M practice in Android world. We then propose an approach that adapts the Rocks toolkit to allow one to rapidly and reliably build complete Android environments (firmware flashes) at the individual sensor level and extend to a large networks of diverse sensors.

DPillar: Dual-port server interconnection network for large scale data centers

To meet the huge demands of computation power and storage space, a future data center may have to include up to millions of servers. The conventional hierarchical tree-based data center network architecture faces several challenges in scaling a data center to that size. Previous research effort has shown that a server-centric architecture, where servers are not only computation and storage workstations but also intermediate nodes relaying traffic for other servers, performs well in scaling a data center to a huge number of servers. This paper presents a server-centric data center network called DPillar, whose topology is inspired by the classic butterfly network. DPillar provides several nice properties and achieves the balance between topological scalability, network performance, and cost efficiency, which make it suitable for building large scale future data centers. Using only commodity hardware, a DPillar network can easily accommodate millions of servers. The structure of a DPillar network is symmetric so that any network bottleneck is eliminated at the architectural level. With each server having only two ports, DPillar is able to provide the bandwidth to support communication intensive distributed applications. This paper studies the interconnection features of DPillar, how to compute routes in DPillar, and how to forward packets in DPillar. Extensive simulation experiments have been performed to evaluate the performance of DPillar. The results show that DPillar performs well even in the presence of a large number of server and switch failures.