Data centers are large scale, energy-hungry infrastructure serving the increasing computational demands as the world is becoming more connected in smart cities. The emergence of advanced technologies such as cloud-based services, internet of things (IoT) and big data analytics has augmented the growth of global data centers, leading to high energy consumption. This upsurge in energy consumption of the data centers not only incurs the issue of surging high cost (operational and maintenance) but also has an adverse effect on the environment. Dynamic power management in a data center environment requires the cognizance of the correlation between the system and hardware level performance counters and the power consumption. Power consumption modeling exhibits this correlation and is crucial in designing energy-efficient optimization strategies based on resource utilization. Several works in power modeling are proposed and used in the literature. However, these power models have been evaluated using different benchmarking applications, power measurement techniques and error calculation formula on different machines. In this work, we present a taxonomy and evaluation of 24 software-based power models using a unified environment, benchmarking applications, power measurement technique and error formula, with the aim of achieving an objective comparison. We use different servers architectures to assess the impact of heterogeneity on the models' comparison. The performance analysis of these models is elaborated in the paper.
Serverless computing has emerged as an attractive deployment option for cloud applications in recent times. The unique features of this computing model include, rapid auto-scaling, strong isolation, fine-grained billing options and access to a massive service ecosystem which autonomously handles resource management decisions. This model is increasingly being explored for deployments in geographically distributed edge and fog computing networks as well, due to these characteristics. Effective management of computing resources has always gained a lot of attention among researchers. The need to a…
Traditionally, the development of computing systems has been focused on performance improvements driven by the demand of applications from consumer, scientific and business domains. However, the ever increasing energy consumption of computing systems has started to limit further performance growth due to overwhelming electricity bills and carbon dioxide footprints. Therefore, the goal of the computer system design has been shifted to power and energy efficiency. To identify open challenges in the area and facilitate future advancements it is essential to synthesize and classify the research on…
Routing in Wireless Sensor Network (WSN) aims to interconnect sensor nodes via single or multi-hop paths. The routes are established to forward data packets from sensor nodes to the sink. Establishing a single path to report each data packet results in increasing energy consumption in WSN, hence, data aggregation routing is used to combine data packets and consequently reduce the number of transmissions. This reduces the routing overhead by eliminating redundant and meaningless data. There are two models for data aggregation routing in WSN: mobile agent and client/server. This paper describes …
Cloud computing based systems, that span data centers, are commonly deployed to offer high performance for user service requests. As data centers continue to expand, computer architects and system designers are facing many challenges on how to balance resource utilization efficiency, server and network performance, energy consumption and quality-of-service (QoS) demands from the users. To develop effective data center management policies, it becomes essential to have an in-depth understanding and synergistic control of the various sub-components inside large scale computing systems, that inclu…
Dynamic Voltage and Frequency Scaling (DVFS), CPU pinning, horizontal, and vertical scaling, are four techniques that have been proposed as actuators to control the performance and energy consumption on data center servers. This work investigates the utility of these four actuators, and quantifies the power-performance tradeoffs associated with them. Using replicas of the German Wikipedia running on our local testbed, we perform a set of experiments to quantify the influence of DVFS, vertical and horizontal scaling, and CPU pinning on end-to-end response time (average and tail), throughput, an…
In this paper, we utilize Mixed Integer Linear Programming (MILP) models to compare the energy efficiency and performance of a server-centric Passive Optical Networks (PON)-based data centers design with different data centers networking topologies for the use in fog computing. For representative MapReduce workloads, completion time results indicate that the server-centric PON-based design achieves 67% reduction in the energy consumption compared to DCell with equivalent performance.
Data centers form a key part of the infrastructure upon which a variety of information technology services are built. They provide the capabilities of centralized repository for storage, management, networking and dissemination of data. With the rapid increase in the capacity and size of data centers, there is a continuous increase in the demand for energy consumption. These data centers not only consume a tremendous amount of energy but are riddled with IT inefficiencies. Data center are plagued with thousands of servers as major components. These servers consume huge energy without performin…
Server consolidation based on virtualization technology simplifies system administration and improves energy efficiency by improving resource utilizations and reducing the physical machine (PM) number in contemporary service-oriented data centers. The elasticity of Internet applications changes the consolidation technologies from addressing virtual machines (VMs) to PMs mapping schemes which must know the VMs statuses, i.e. the number of VMs and the profiling data of each VM, into providing the application-to-VM-to-PM mapping. In this paper, we study on the consolidation of multiple Internet a…
Data center providers seek to minimize their total cost of ownership (TCO), while power consumption has become a social concern. We present formulations to minimize server energy consumption and server cost under three different data center server setups (homogeneous, heterogeneous, and hybrid hetero-homogeneous clusters) with dynamic temporal workload. Our studies show that the homogeneous model significantly differs from the heterogeneous model in computational time (by an order of magnitude). To be able to compute optimal configurations in near real-time for large scale data centers, we pro…
Fog computing was designed to support the specific needs of latency-critical applications such as augmented reality, and IoT applications which produce massive volumes of data that are impractical to send to faraway cloud data centers for analysis. However this also created new opportunities for a wider range of applications which in turn impose their own requirements on future fog computing platforms. This article presents a study of a representative set of 30 fog computing applications and the requirements that a general-purpose fog computing platform should support.
Web application providers have been migrating their applications to cloud data centers, attracted by the emerging cloud computing paradigm. One of the appealing features of the cloud is elasticity. It allows cloud users to acquire or release computing resources on-demand, which enables web application providers to automatically scale the resources provisioned to their applications without human intervention under a dynamic workload to minimize resource cost while satisfying Quality of Service (QoS) requirements. In this paper, we comprehensively analyze the challenges that remain in auto-scali…
This paper considers a cost minimization problem for data centers with N servers and randomly arriving service requests. A central router decides which server to use for each new request. Each server has three types of states (active, idle, setup) with different costs and time durations. The servers operate asynchronously over their own states and can choose one of multiple sleep modes when idle. We develop an online distributed control algorithm so that each server makes its own decisions, the request queues are bounded and the overall time average cost is near optimal with probability 1. The…
The recently proposed RCube network is a cube-based server-centric data center network (DCN), including two types of heterogeneous servers, called core and edge servers. Remarkably, it takes the latter as backup servers to deal with server failures and thus achieve high availability. This paper first points out that RCube is suitable as a candidate topology of DCNs for edge computing. Three transmission types are among core and edge servers based on the demand for applications' computation and instant response. We then employ protection routing to analyze the transmission failure of RCube DCNs…
Energy consumption is a growing issue in data centers, impacting their economic viability and their public image. In this work we empirically characterize the power and energy consumed by different types of servers. In particular, in order to understand the behavior of their energy and power consumption, we perform measurements in different servers. In each of them, we exhaustively measure the power consumed by the CPU, the disk, and the network interface under different configurations, identifying the optimal operational levels. One interesting conclusion of our study is that the curve that d…
ARM processors have dominated the mobile device market in the last decade due to their favorable computing to energy ratio. In this age of Cloud data centers and Big Data analytics, the focus is increasingly on power efficient processing, rather than just high throughput computing. ARM's first commodity server-grade processor is the recent AMD A1100-series processor, based on a 64-bit ARM Cortex A57 architecture. In this paper, we study the performance and energy efficiency of a server based on this ARM64 CPU, relative to a comparable server running an AMD Opteron 3300-series x64 CPU, for Big …
Parallel computing is the fundamental base for MapReduce framework in Hadoop. Each data chunk is replicated over 3 servers for increasing availability of data and decreasing probability of data loss. Hence, the 3 servers that have Map task stored on their disk are fastest servers to process them, which are called local servers. All servers in the same rack as local servers are called rack-local servers that are slower than local servers since data chunk associated with Map task should be fetched through top of the rack switch. All other servers are called remote servers that are slowest server…
Cloud Data Centers have become the backbone infrastructure to provide services. With the emerging edge computing paradigm, computation and networking capabilities have been pushed from clouds to the edge to provide computation, intelligence, networking management with low end-to-end latency. Service migration across different computing nodes in edge and cloud computing becomes essential to guarantee the quality of service in the dynamic environment. Many studies have been conducted on the dynamic resource management involving migrating Virtual Machines to achieve various objectives, such as lo…
Users of cloud computing platforms pose different types of demands for multiple resources on servers (physical or virtual machines). Besides differences in their resource capacities, servers may be additionally heterogeneous in their ability to service users - certain users' tasks may only be serviced by a subset of the servers. We identify important shortcomings in existing multi-resource fair allocation mechanisms - Dominant Resource Fairness (DRF) and its follow up work - when used in such environments. We develop a new fair allocation mechanism called Per-Server Dominant-Share Fairness (PS…
Massive data centers housing thousands of computing nodes have become commonplace in enterprise computing, and the power consumption of such data centers is growing at an unprecedented rate. Adding to the problem is the inability of the servers to exhibit energy proportionality, i.e., provide energy-efficient execution under all levels of utilization, which diminishes the overall energy efficiency of the data center. It is imperative that we realize effective strategies to control the power consumption of the server and improve the energy efficiency of data centers. With the advent of Intel Sa…
Containers, enabling lightweight environment and performance isolation, fast and flexible deployment, and fine-grained resource sharing, have gained popularity in better application management and deployment in addition to hardware virtualization. They are being widely used by organizations to deploy their increasingly diverse workloads derived from modern-day applications such as web services, big data, and IoT in either proprietary clusters or private and public cloud data centers. This has led to the emergence of container orchestration platforms, which are designed to manage the deployment…
We study the dynamic power optimization problem in data centers. We formulate and solve the following offline problem: in which slot which server has to be assigned to which job; and in which slot which server has to be switched ON or OFF so that the total power is optimal for some time horizon. We show that the offline problem is a new version of generalized assignment problem including new constraints issuing from deadline characteristics of jobs and difference of activation energy of servers. We propose an online algorithm that solves the problem heuristically and compare it to randomized r…
University of California Merced（加州大学默塞德分校）；San Jose State University（圣何塞州立大学）
With the growing complexity of big data workloads that require abundant data and computation, data centers consume a tremendous amount of power daily. In an effort to minimize data center power consumption, several studies developed power models that can be used for job scheduling either reducing the number of active servers or balancing workloads across servers at their peak energy efficiency points. Due to increasing software and hardware heterogeneity, we observed that there is no single power model that works the best for all server conditions. Some complicated machine learning models them…
Cloud computing is being viewed as the technology of today and the future. Through this paradigm, the customers gain access to shared computing resources located in remote data centers that are hosted by cloud providers (CP). This technology allows for provisioning of various resources such as virtual machines (VM), physical machines, processors, memory, network, storage and software as per the needs of customers. Application providers (AP), who are customers of the CP, deploy applications on the cloud infrastructure and then these applications are used by the end-users. To meet the fluctuatin…
In Wolke et al. [1] we compare the efficiency of different resource allocation strategies experimentally. We focused on dynamic environments where virtual machines need to be allocated and deallocated to servers over time. In this companion paper, we describe the simulation framework and how to run simulations to replicate experiments or run new experiments within the framework.
Nowadays cloud computing adoption as a form of hosted application and services is widespread due to decreasing costs of hardware, software, and maintenance. Cloud enables access to a shared pool of virtual resources hosted in large energy-hungry data centers for diverse information and communication services with dynamic workloads. The huge energy consumption of cloud data centers results in high electricity bills as well as emission of a large amount of carbon dioxide gas. Needless to say, efficient resource management in cloud environments has become one of the most important priorities of c…
In this paper we present new algorithmic solutions for several constrained geometric server placement problems. We consider the problems of computing the 1-center and obnoxious 1-center of a set of line segments, constrained to lie on a line segment, and the problem of computing the K-median of a set of points, constrained to lie on a line. The presented algorithms have applications in many types of distributed systems, as well as in various fields which make use of distributed systems for running some of their applications (like chemistry, metallurgy, physics, etc.).
Today's datacenters face important challenges for providing low-latency high-quality interactive services to meet user's expectation. For improving the application throughput, recent research works have embedded application deadline information into design of network flow schedule to meet the latency requirement. Here, arises a critical question: does application-level throughput mean providing better quality service? We note that there are usually a set of semantic related responses (or flows) for answering a query; and, some responses are highly correlative with the query while others do not…
In cloud data center (CDC), reducing energy consumption while maintaining performance has always been a hot issue. In server consolidation, the traditional solution is to divide the problem into multiple small problems such as host overloading detection, virtual machine (VM) selection and VM placement and solve them step by step. However, the design of host overloading detection strategies and VM selection strategies cannot be directly linked to the ultimate goal of reducing energy consumption and ensuring performance. This paper proposes a learning-based VM selection strategy that selects app…
Internet-scale distributed systems such as content delivery networks (CDNs) operate hundreds of thousands of servers deployed in thousands of data center locations around the globe. Since the energy costs of operating such a large IT infrastructure are a significant fraction of the total operating costs, we argue for redesigning CDNs to incorporate energy optimizations as a first-order principle. We propose techniques to turn off CDN servers during periods of low load while seeking to balance three key design goals: maximize energy reduction, minimize the impact on client-perceived service ava…
For a provider of a Content Delivery Network (CDN), the location selection of mirror servers is a complex optimization problem. Generally, the objective is to place the nodes centralized such that all customers have convenient access to the service according to their demands. It is an instance of the k-center problem, which is proven to be NP-hard. Determining reasonable server locations directly influences run time effects and future service costs. We model, simulate, and optimize the properties of a content delivery network. Specifically, considering the server locations in a network infrast…
Data centers are becoming increasingly popular for their flexibility and processing capabilities in the modern computing environment. They are managed by a single entity (administrator) and allow dynamic resource provisioning, performance optimization as well as efficient utilization of available resources. Each data center consists of massive compute, network and storage resources connected with physical wires. The large scale nature of data centers requires careful planning of compute, storage, network nodes, interconnection as well as inter-communication for their effective and efficient op…
An increasing number of technology enterprises are adopting cloud-native architectures to offer their web-based products, by moving away from privately-owned data-centers and relying exclusively on cloud service providers. As a result, cloud vendors have lately increased, along with the estimated annual revenue they share. However, in the process of selecting a provider's cloud service over the competition, we observe a lack of universal common ground in terms of terminology, functionality of services and billing models. This is an important gap especially under the new reality of the industry…
ELLIS Alicante（ELLIS阿利坎特研究所）；University of Alicante（阿利坎特大学）；University of Sussex（萨塞克斯大学）
Federated learning (FL) has been proposed as a privacy-preserving approach in distributed machine learning. A federated learning architecture consists of a central server and a number of clients that have access to private, potentially sensitive data. Clients are able to keep their data in their local machines and only share their locally trained model's parameters with a central server that manages the collaborative learning process. FL has delivered promising results in real-life scenarios, such as healthcare, energy, and finance. However, when the number of participating clients is large, t…
In this paper, we utilize Mixed Integer Linear Programming (MILP) models to determine the impact of link failures on the performance of shuffling operations in MapReduce when different data center network (DCN) topologies are used. For a set of non-fatal single and multi-links failures, the results indicate that different DCNs experience different completion time degradations ranging between 5% and 40%. The best performance under links failures is achieved by a server-centric PON-based DCN.
Virtual machine (VM) placement is very important for cloud platforms. While techniques, such as live virtual machine migration, are very useful to balance the load in the data centers, they are expensive operations. In this position paper, we propose to minimize the chance of the load hot spots in the data center by applying the workload patterns of the VMs in the virtual machine placement algorithms - place VMs that require a lot of same type of resource across different physical servers. In this way, the resource competition of VMs on the same physical server is significantly mitigated. Mean…
Nowadays, deep neural networks (DNNs) are the core enablers for many emerging edge AI applications. Conventional approaches to training DNNs are generally implemented at central servers or cloud centers for centralized learning, which is typically time-consuming and resource-demanding due to the transmission of a large amount of data samples from the device to the remote cloud. To overcome these disadvantages, we consider accelerating the learning process of DNNs on the Mobile-Edge-Cloud Computing (MECC) paradigm. In this paper, we propose HierTrain, a hierarchical edge AI learning framework, …
