Data Center Hardware Research Articles

Holistic Resource Scheduling for Data Center In-Network Computing

The recent trend towards more programmable switching hardware in data centers opens up new possibilities for distributed applications to leverage in-network computing (INC). Literature so far has largely focused on individual application scenarios of INC, leaving aside the problem of coordinating usage of potentially scarce and heterogeneous switch resources among multiple INC scenarios, applications, and users. Alas, the traditional model of resource pools of isolated compute containers does not fit an INC-enabled data center. This paper describes HIRE, a holistic INC-aware resource manager which allows for server-local and INC resources to be coordinated in unison. HIRE introduces a novel flexible resource (meta-)model to address heterogeneity and resource interchangeability, and includes two approaches for INC scheduling: (a) retrofitting existing schedulers; (b) designing a new one. For (a), HIRE presents a retrofitting API and demonstrates it with four state-of-the-art schedulers. For (b), HIRE proposes a flow-based scheduler, cast as a min-cost max-flow problem, where a unified cost model is used to integrate the different costs. Experiments with a workload trace of a 4000 machine cluster show that HIRE makes better use of INC resources by serving 8–30% more INC requests, while simultaneously reducing network detours by 20% and reducing tail placement latency by 50%.

Візуалізація процесів пароутворення у двофазному термосифоні за різної його орієнтації в просторі

Currently, thermosyphons are used to cool such devices as power amplifiers of radio frequency systems, data center hardware, LED light sources, etc. One of the important factors affecting the efficiency of such cooling systems is the orientation of the thermosyphons in space. This paper is dedicated to research and visualization of vaporization events in two-phase thermosyphons, primarily focusing on investigating and visualizing the influence of orientation in space on vaporization. The studies were performed for100% fill ratio. Vaporization was video recorded at 240 frames per second, whereupon the obtained footage was converted into image sequence. The analysis of the obtained materials has shown that at intensive boiling for tilt angle range of 5—45º, a part of the working fluid is always outside the evaporator. When the angle is decreased, the portion of the working fluid outside the evaporator increases. Moreover, for the 5° tilt angle, the evaporator can be completely drained at certain time intervals. It is proposed that the thermal resistance at low heat flux values for smaller tilt angles can be lower than for vertical orientation because of thin liquid films. Evaporation is more effective in thin liquid films than in large volumes. This prompts the conclusion that smaller angles will allow obtaining lower thermal resistance for the same filling ratio. On the other hand, maximum heat transfer ability decreases dramatically for the angle range of 0—10º. The obtained results can be used in further studies to analyze and explane the aspects of heat transfer in two-phase thermosyphons with a short evaporator.

Virtual machine optimization to achieve energy efficient optimum resource utilization in cloud data center

Objectives: Cloud offers multiple benefits through its data center-based services. The whole world uses these services that are hosted by physical machines. Millions of virtual machines get an optimized utilization of hardware to utilize these physical machines. The unbalanced distribution of virtual machines to physical machines offers an in-efficient utilization of data center hardware that will also lead to more carbon emission and harm the environment as well. Methods/material: A learning function is needed to offer energy efficiency in the cloud data center through VM optimization through its optimal allocation to physical machines. Therefore, an optimal VM placement and migration algorithm is a challenge that is addressed in this paper to reach an efficient energy optimization and resource utilization level. Finding/Novelty: The proposed algorithm is led by a learning function that takes into account the available number of physical machines, number of virtual machines, incoming requests and decides to run an optimal number of physical machines to obtain energy efficiency level for the cloud data center by migrating the virtual machines (VMs). Keywords: Datacenter; Cloud computing; Energy efficiency; VM placement; VM migration; Green Cloud

On the classification and quantification of server consolidation overheads

Virtualization technology is important for servers that compose cloud data centers. The current trend is to consolidate servers to manage them easily and reduce hardware and power consumption in data centers. However, performance degradation is inherent to virtualization technology and is caused by the hypervisor and overhead due to the consolidation of several virtual servers inside a physical server. Several ways exist to virtualize a server; these methods are based mainly on virtual machines and containers. In this paper, we propose a general method to estimate the value of the consolidation overhead classes, regardless of the virtualization platform, server characteristics and workload type. We conducted several experiments in different scenarios to illustrate the usefulness of the proposed method. The results show the applicability of the proposed method and indicate that these inherent overheads are not negligible in many cases depending on, first, the type of hypervisor and, second, the hardware resources features of the physical server.

Opening up the Baseboard Management Controller

In 2011 Facebook announced the Open Compute Project to form a community around open-source designs and specifications for data center hardware. Since then, the project has expanded to all aspects of the open data center. This column focuses on the BMC and is an introduction to a complicated topic. The intention is to provide a full picture of the world of the open-source BMC ecosystem, starting with a brief overview of the BMC’s role in a system, touching on security concerns around the BMC, and then diving into some of the projects that have developed in the open-source ecosystem.

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.

Hybrid Electro-Optical Intra-Data Center Networks Tailored for Different Traffic Classes

This work presents a novel approach in intradata center network design and performance evaluation, based on a tailored, long-range dependence data traffic generation model for different application classes. We examine how an intra-data center network can be efficiently evaluated and optimized by applying accurate models for machine-generated data and using assumptions on current and future commercially available data center hardware. Moreover, we show that by migrating such a Fat-Tree network to its hybrid, electro-optical counterpart by employing optical circuit switching for selected, intense traffic connections, significant capital and operational expenditure cost savings can be obtained.

Decision-making and management system based on internet of things for Hulun Buir State Farm

According to the specific requirements of scientific decision-making and precision production control in the production process of modern grasslands and animal husbandry, this paper presents a scheme of production command decision system of modern grasslands and animal husbandry based on internet of things. The scheme is composed of four major components, including big data and cloud computing, grassland animal husbandry production management decision information system, precision operating technology and equipment systems, e-commerce and full traceability system. Specific description is as follows: (ⅰ) Big data and cloud computing. According to the large-scale and high-value characteristics of big data, theoretical method and technical system of big data acquisition, management and services of grass and animal husbandry are perfectly established. Large data center hardware platform is constructed based on cloud computing technology. PB(Petabyte) level grassland animal husbandry data resources are integrated and large data intelligent services are carried out based on agricultural vertical search engine, agricultural cloud data management, agricultural knowledge ontology and other key technologies. (ⅱ) Grassland animal husbandry production management decision information system. According to the organizational structure and production management model of Hulun Buir State Farm, the agricultural production management and decision information system of Hulun Buir is developed in accordance with the concept of “horizontal to the side, vertical to the end”, that is, horizontal to edge refers to the data information that can be obtained from each group, sub-bureau, farm user and the production team user; vertical to bottom refers to the latest information that can be obtained from every field crop, every farm machine, and every cow. A data warehouse is built, including comparing current accumulated temperature data and rainfall data with over years. The monitoring data of soil, including soil moisture and temperature, conductivity and salt, provide accurate soil parameters for the grass and animal husbandry, and the installation of video surveillance points timely access to soil information. (ⅲ) Precision operating technology and equipment systems. Based on herbage yield information, satellite and unmanned aerial vehicle remote sensing data and measured data of farmland soil nutrients, the spatial variability information of farmland fertility is obtained, and the fertilizer prescription map is formulated. In the process of operation, the GPS data are acquired in real time and the data of fertilizer prescription maps are read, and the accurate fertilization operation is realized through the variable fertilization control system and the equipment. (ⅳ) E-commerce and full traceability system. Business electronic platform of grass industry can perfectly provide the business service to the merchants. Each batch of grass and derivatives will be able to ensure quality, quantity, and customer interests through traceability systems in the future, so that production has records, flows can be traced and responsibility can be defined. By content of decision system, low level of production mode, low degree of intelligent and fine control and lack of market information guidance of production platform are solved in large and medium-sized grass pasture in China, in order to enhance the level of modern grass animal husbandry information and help modern grass production and transformation of the production mode.

A Review on Attribute Based Encryption

Cloud computing is an emerging technology. It is a basic paradigm which satisfies all the needs of a customer and it may even serve as a fifth utility in near future. In general it refers to both the applications delivered as services over the Internet and the hardware and systems software in the data centers that provide those services. The data center hardware and software is known as a cloud. A huge amount of data is being processed and shared. So, while we are sharing a large amount of data through online there is a chance for the data to get modified by someone else. In order to protect this data from getting modified we provide several security mechanism like access control. One of the mechanism would be by encrypting and sending the data. Encryption plays a major role in protecting the data. Many encryption schemes were proposed but it cannot survive long due to efficiency constraint. So firstly attribute based encryption was proposed which was later modified to key policy based encryption scheme which was again modified by cipher text encryption scheme is and was later again switched to file hierarchy cipher text based encryption. So my project mainly concentrates on File Hierarchy Attribute Based Encryption scheme in which we make use of Hierarchical structures and attributes matching which is mainly important for providing access control to the data.

Placement and Performance Analysis of Virtual Multicast Networks in Fat-Tree Data Center Networks

Virtualization of servers and networks is a key technique to resolve the conflict between the increasing demands on computing power and the high cost of hardware in data centers. In order to map virtual networks to physical infrastructure efficiently, designers have to make careful decisions on the allocation of limited resources, which makes placement of virtual networks in data centers a critical issue. In this paper, we study the placement of virtual networks in fat-tree data center networks. In order to meet the requirements of instant parallel data transfer between multiple computing units, we propose a model of multicast-capable virtual networks (MVNs). We then design four virtual machine (VM) placement schemes to embed MVNs into fat-tree data center networks, named Most-Vacant-Fit (MVF), Most-Compact-First (MCF), Mixed-Bidirectional-Fill (MBF), and Malleable-Shallow-Fill (MSF). All these VM placement schemes guarantee the nonblocking multicast capability of each MVN while simultaneously achieving significant saving in the cost of network hardware. In addition, each VM placement scheme has its unique features. The MVF scheme has zero interference to existing computing tasks in data centers; the MCF scheme leads to the greatest cost saving; the MBF scheme simultaneously possesses the merits of MVF and MCF, and it provides an adjustable parameter allowing cloud providers to achieve preferred balance between the cost and the overhead; the MSF scheme performs at least as well as MBF, and possesses some additional predictable features. Finally, we compare the performance and overhead of these VM placement schemes, and present simulation results to validate the theoretical results.

Implementation and performance analysis of various VM placement strategies in CloudSim

Infrastructure as a Service (IaaS) has become one of the most dominant features that cloud computing offers nowadays. IaaS enables datacenter's hardware to get virtualized which allows Cloud providers to create multiple Virtual Machine (VM) instances on a single physical machine, thus improving resource utilization and increasing the Return on Investment (ROI). VM consolidation includes issues like choosing appropriate algorithm for selection of VMs for migration and placement of VMs to suitable hosts. VMs need to be migrated from overutilized host to guarantee that demand for computer resources and performance requirements are accomplished. Besides, they need to be migrated from underutilized host to deactivate that host for saving power consumption. In order to solve the problem of energy and performance, efficient dynamic VM consolidation approach is introduced in literature. In this work, we have proposed multiple redesigned VM placement algorithms and introduced a technique by clustering VMs to migrate by taking account both CPU utilization and allocated RAM. We implement and study the performance of our algorithms on a cloud computing simulation toolkit known as CloudSim using PlanetLab workload data. Simulation results demonstrate that our proposed techniques outperform the default VM Placement algorithm designed in CloudSim.

Improving MapReduce performance by balancing skewed loads

MapReduce has emerged as a popular computing model used in datacenters to process large amount of datasets. In the map phase, hash partitioning is employed to distribute data that sharing the same key across data center-scale cluster nodes. However, we observe that this approach can lead to uneven data distribution, which can result in skewed loads among reduce tasks, thus hamper performance of MapReduce systems. Moreover, worker nodes in MapReduce systems may differ in computing capability due to (1) multiple generations of hardware in non-virtualized data centers, or (2) co-location of virtual machines in virtualized data centers. The heterogeneity among cluster nodes exacerbates the negative effects of uneven data distribution. To improve MapReduce performance in heterogeneous clusters, we propose a novel load balancing approach in the reduce phase. This approach consists of two components: (1) performance prediction for reducers that run on heterogeneous nodes based on support vector machines models, and (2) heterogeneity-aware partitioning (HAP), which balances skewed data for reduce tasks. We implement this approach as a plug-in in current MapReduce system. Experimental results demonstrate that our proposed approach distributes work evenly among reduce tasks, and improves MapReduce performance with little overhead.

Power-Saving Management for Energy-Efficient Green Clouds

How to efficiently utilize energy has become one of the most important issues in green computing. Information technology has been employing the ability of technology to optimize resource utilization resulted in less waste of resources. As the era of information technology get into the cloud computing technology, applications are delivered as services over the Internet and the hardware and systems software in the data centers provide those services. Cloud technology might hold promise in reducing energy consumption and greenhouse effect. In the same time, the cumulative depletion of energy for the need of cloud operation is huge, especially in the large computational and data center. These servers should be controlled and used efficiently for saving energy. Therefore, in this paper we present an energy management strategy of cloud. Using proposed calculation model of developed method to reduce the numbers of wake up server, we may find the way to allocate or focus on the workload of server in higher operating efficiency. And that will achieve better energy saving.

Global cost diversity aware dispatch algorithm for heterogeneous data centers (abstracts only)

Large, Internet based companies service user requests from multiple data centers located across the globe. These data centers often house a heterogeneous computing infrastructure and draw electricity from the local electricity market.Reducing the electricity costs of operating these data centers is a challenging problem, and in this work, we propose a novel solution which exploits both the data center heterogeneity and global electricity market diversity to reduce data center operating cost. We evaluate our solution in our test-bed that simulates a heterogeneous data center, using real-world request workload and real-world electricity prices. We show that our strategies achieve cost and energy saving of at least 21% over a naive load balancing scheme that distributes requests evenly across data centers, and outperform existing solutions which either do not exploit the electricity market diversity or do not exploit data center hardware diversity.

Optimizing thermal design of data center cabinets with a new multi-objective genetic algorithm

There is an ever increasing need to use optimization methods for thermal design of data centers and the hardware populating them. Airflow simulations of cabinets and data centers are computationally intensive and this problem is exacerbated when the simulation model is integrated with a design optimization method. Generally speaking, thermal design of data center hardware can be posed as a constrained multi-objective optimization problem. A popular approach for solving this kind of problem is to use Multi-Objective Genetic Algorithms (MOGAs). However, the large number of simulation evaluations needed for MOGAs has been preventing their applications to realistic engineering design problems. In this paper, details of a substantially more efficient MOGA are formulated and demonstrated through a thermal analysis simulation model of a data center cabinet. First, a reduced-order model of the cabinet problem is constructed using the Proper Orthogonal Decomposition (POD). The POD model is then used to form the objective and constraint functions of an optimization model. Next, this optimization model is integrated with the new MOGA. The new MOGA uses a "kriging" guided operation in addition to conventional genetic algorithm operations to search the design space for global optimal design solutions. This approach for optimal design is essential to handle complex multi-objective situations, where the optimal solutions may be non-obvious from simple analyses or intuition. It is shown that in optimizing the data center cabinet problem, the new MOGA outperforms a conventional MOGA by estimating the Pareto front using 50% fewer simulation calls, which makes its use very promising for complex thermal design problems.