SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
This case study compares energy and usability of remote virtual machines accessed through zero-client devices with traditional laptop computers to determine if zero-client computing could be a solution to reducing building energy use.
A network of virtual machines at cloud server sites connected over virtual IO connections is a flexible, easily deployable, and cost-effective alternative to a physical network infrastructure with dedicated servers connected over leased fiber lines. We study the throughput performance of such a cloud network by collecting measurements over Google Cloud infrastructure spanning multiple continents. To study its ideal performance and impact of packet losses, we utilize its emulation using dedicated servers and connection hardware emulation devices. We compare the measurements over the cloud network to those over its emulation on a testbed. We examine the throughput profiles of both networks as a function of the round trip time and their utilization-concavity coefficients, estimated using measurements for common TCP versions. The throughput profile's concave-convex shape and its coefficient are critical indicators of the network performance, qualitatively and quantitatively, respectively. The results indicate their overall agreement between the production cloud network and its emulation using dedicated connections, and a near optimal throughput performance of the former except for a few under-performing connections. Also, the number of parallel flows is found to be a dominant factor in optimizing the throughput across various conditions and TCP versions.
In prior work (Yoginath and Perumalla, 2011; Yoginath, Perumalla and Henz, 2012), the motivation, challenges and issues were articulated in favor of virtual time ordering of Virtual Machines (VMs) in network simulations hosted on multi-core machines. Two major components in the overall virtualization challenge are (1) virtual timeline establishment and scheduling of VMs, and (2) virtualization of inter-VM communication. Here, we extend prior work by presenting scaling results for the first component, with experiment results on up to 128 VMs scheduled in virtual time order on a single 12-core host. We also explore the solution space of design alternatives for the second component, and present performance results from a multi-threaded, multi-queue implementation of inter-VM network control for synchronized execution with VM scheduling, incorporated in our NetWarp simulation system.
Synchrophasor technology plays a pivotal role in developing the next generation of wide-area monitoring, protection, and control in the smart grid environment. As technology and communications infrastructures evolve, however, so do the attack surfaces in the synchrophasor network that can be exploited by advanced persistent threat (APT) actors to affect power system stability and reliability. In this paper, we propose a novel platform for developing a self-healing and attack-resilient PMU network (SHARP-Net) by instituting a state-of-the-art intrusion detection system (IDS) with an intrusion mitigation system (IMS) and an alert management system (AMS). In particular, the proposed platform detects anomalies during cyberattacks on phasor data concentrators (PDCs) based on the rules defined in the IDS, then the generated alerts are published to the IMS through the AMS. The proposed IMS proceeds to take automated corrective responses to mitigate cyberattacks by reconfiguring the synchrophasor network to isolate the compromised PDCs, and it orchestrates new PDCs to prevent the future propagation of attacks. Further, the IMS restores the system's observability by reconnecting the new PDCs to make the grid attack-resilient. In this work, the SHARP-Net platform is developed by using Python-based libraries, minimega's software-defined network, and virtual machine orchestration. We implement and validate the proposed SHARP-Net architecture by testing a PMU network in the smart grid environment. SHARP-Net showed promising performance in detecting cyberattacks and mitigating them through the network reconfiguration.
Plug loads account for approximately 30% of commercial whole-building energy use, and this percentage is growing. That means the United States spends about $60 billion per year on energy to keep our devices running. About 10% of this is attributed to computing resources, which includes both traditional laptop and desktop computers, as well as data centers that can host virtual machines (VMs) and enable teleworking trends. Teleworking and more flexible work environments are becoming increasingly popular options, and many business and building owners wonder whether hosting VMs on data center servers has the potential to save energy. To answer this question, researchers at the National Renewable Energy Laboratory (NREL) conducted a study comparing energy and usability of remote VMs (accessed through zero-client devices) with traditional laptop computers. The findings were published in the Journal of Sustainable Cities and Society in 2017.
Self-excited induction generators offer a robust solution for power production for standalone as well as grid-connected systems. In general, self-excited induction generators require excitation capacitors which make use of the machine magnetization characteristics for voltage build up process as well as operation at a specific frequency. In this paper, a self-excited induction machine is modeled with both the electrical and mechanical dynamics. This modeled virtual machine's dynamics are utilized for voltage build up process for a standalone photovoltaic converter connected to a local load for a microgrid application. The modeled machine's parameters are used from the name plate rating from the manufacturer. However, in a microgrid the accommodation of unbalanced and/or nonlinear harmonic rich load is a necessity, therefore, in this work the virtual self-excitation capacitors of the modeled machine are varied based on the machine characteristics. With the objective of ensuring harmonic free point of common coupling voltage, the modeled virtual self-excitation capacitors are varied to accomplish change in terminal frequency and the virtual load torque is varied to obtain voltage magnitude change. To verify the efficacy, the overall system is modeled in MATLAB/Simulink and PLECS domain and most important case studies are presented.
This study provides a detailed comparison of the power consumption, usability, and applicability of virtual machines (VMs) - accessed through zero-client devices - and traditional laptop computers in office settings. The study analyzed high-level plug loads across two office spaces, one using traditional laptops, the other using zero clients. In addition, the individual power consumption of four workstations was monitored and compared. Each workstation user switched between using a laptop or zero client for one week and then using the alternate system the second week. Results of the high-level and workstation analysis show that average workstation plug loads are lower for occupants using zero clients. However, this does not include power consumed by the data center managing VMs. This study calculates the affiliated data center power draw of VMs and shows that server-related loads push total zero-client computing energy higher than that of traditional laptops. Finally, a questionnaire was administered to building occupants to determine the appropriateness of zero-client computing for various user groups, with the results indicating that VMs are most appropriate for more basic software functions. The findings of this study suggest that advances in server technology can help improve both the overall efficiency and the usability of zero-client computing.
The Fermilab Grid and Cloud Computing Department and the KISTI Global Science experimental Data hub Center are working on a multi-year Collaborative Research and Development Agreement.With the knowledge developed in the first year on how to provision and manage a federation of virtual machines through Cloud management systems. In this second year, we expanded the work on provisioning and federation, increasing both scale and diversity of solutions, and we started to build on-demand services on the established fabric, introducing the paradigm of Platform as a Service to assist with the execution of scientific workflows. We have enabled scientific workflows of stakeholders to run on multiple cloud resources at the scale of 1,000 concurrent machines. The demonstrations have been in the areas of (a) Virtual Infrastructure Automation and Provisioning, (b) Interoperability and Federation of Cloud Resources, and (c) On-demand Services for ScientificWorkflows.
Disclosed in some examples are methods, systems, memory devices, memory controllers, and machine-readable mediums which provide for reserving physical memory device resources to specific execution units. Execution units may include processes, threads, virtual machines, functions, procedures, or the like. Physical memory device resources may include channels, modules, ranks, banks, bank groups, and the like. For example, a physical memory device resource that is reservable may be a smallest unit that allows for parallel access with another of the same size unit.
Fermilab supports a scientific program that includes experiments and scientists located across the globe. To better serve this community, in 2004, the (then) Computing Division undertook the strategy of placing all of the High Throughput Computing (HTC) resources in a Campus Grid known as FermiGrid, supported by common shared services. In 2007, the FermiGrid Services group deployed a service infrastructure that utilized Xen virtualization, LVS network routing and MySQL circular replication to deliver highly available services that offered significant performance, reliability and serviceability improvements. This deployment was further enhanced through the deployment of a distributed redundant network core architecture and the physical distribution of the systems that host the virtual machines across multiple buildings on the Fermilab Campus. In 2010, building on the experience pioneered by FermiGrid in delivering production services in a virtual infrastructure, the Computing Sector commissioned the FermiCloud, General Physics Computing Facility and Virtual Services projects to serve as platforms for support of scientific computing (FermiCloud 6 GPCF) and core computing (Virtual Services). Lastly, this work will present the evolution of the Fermilab Campus Grid, Virtualization and Cloud Computing infrastructure together with plans for the future.
High performance computing environments are often used for a wide variety of workloads ranging from simulation, data transformation and analysis, and complex workflows to name just a few. These systems may process data for a variety of users, often requiring strong separation between job allocations. There are many challenges to establishing these secure enclaves within the shared infrastructure of high-performance computing (HPC) environments. The isolation mechanisms in the system software are the basic building blocks for enabling secure compute enclaves. There are a variety of approaches and the focus of this report is to review the different virtualization technologies that facilitate the creation of secure compute enclaves. The report reviews current operating system (OS) protection mechanisms and modern virtualization technologies to better understand the performance/isolation properties. We also examine the feasibility of running ``virtualized'' computing resources as non-privileged users, and providing controlled administrative permissions for standard users running within a virtualized context. Our examination includes technologies such as Linux containers (LXC [32], Docker [15]) and full virtualization (KVM [26], Xen [5]). We categorize these different approaches to virtualization into two broad groups: OS-level virtualization and system-level virtualization. The OS-level virtualization uses containers to allow a single OS kernel to be partitioned to create Virtual Environments (VE), e.g., LXC. The resources within the host's kernel are only virtualized in the sense of separate namespaces. In contrast, system-level virtualization uses hypervisors to manage multiple OS kernels and virtualize the physical resources (hardware) to create Virtual Machines (VM), e.g., Xen, KVM. This terminology of VE and VM, detailed in Section 2, is used throughout the report to distinguish between the two different approaches to providing virtualized execution environments. As part of our technology review we analyzed several current virtualization solutions to assess their vulnerabilities. This included a review of common vulnerabilities and exposures (CVEs) for Xen, KVM, LXC and Docker to gauge their susceptibility to different attacks. The complete details are provided in Section 5 on page 33. Based on this review we concluded that system-level virtualization solutions have many more vulnerabilities than OS level virtualization solutions. As such, security mechanisms like sVirt (Section 3.3) should be considered when using system-level virtualization solutions in order to protect the host against exploits. The majority of vulnerabilities related to KVM, LXC, and Docker are in specific regions of the system. Therefore, future "zero day attacks" are likely to be in the same regions, which suggests that protecting these areas can simplify the protection of the host and maintain the isolation between users. The evaluations of virtualization technologies done thus far are discussed in Section 4. This includes experiments with 'user' namespaces in VEs, which provides the ability to isolate user privileges and allow a user to run with different UIDs within the container while mapping them to non-privileged UIDs in the host. We have identified Linux namespaces as a promising mechanism to isolate shared resources, while maintaining good performance. In Section 4.1 we describe our tests with LXC as a non-root user and leveraging namespaces to control UID/GID mappings and support controlled sharing of parallel file-systems. We highlight several of these namespace capabilities in Section 6.2.3. The other evaluations that were performed during this initial phase of work provide baseline performance data for comparing VEs and VMs to purely native execution. In Section 4.2 we performed tests using the High-Performance Computing Conjugate Gradient (HPCCG) benchmark to establish baseline performance for a scientific application when run on the Native (host) machine in contrast with execution under Docker and KVM. Our tests verified prior studies showing roughly 2-4% overheads in application execution time & MFlops when running in hypervisor-base environments (VMs) as compared to near native performance with VEs. For more details, see Figures 4.5 (page 28), 4.6 (page 28), and 4.7 (page 29). Additionally, in Section 4.3 we include network measurements for TCP bandwidth performance over the 10GigE interface in our testbed. The Native and Docker based tests achieved >= ~9Gbits/sec, while the KVM configuration only achieved 2.5Gbits/sec (Table 4.6 on page 32). This may be a configuration issue with our KVM installation, and is a point for further testing as we refine the network settings in the testbed. The initial network tests were done using a bridged networking configuration. The report outline is as follows: - Section 1 introduces the report and clarifies the scope of the proj...
The primary goal of the Harness WorkBench (HWB) project is to investigate innovative software environments that will help enhance the overall productivity of applications science on diverse HPC platforms. Two complementary frameworks were designed: one, a virtualized command toolkit for application building, deployment, and execution, that provides a common view across diverse HPC systems, in particular the DOE leadership computing platforms (Cray, IBM, SGI, and clusters); and two, a unified runtime environment that consolidates access to runtime services via an adaptive framework for execution-time and post processing activities. A prototype of the first was developed based on the concept of a 'system-call virtual machine' (SCVM), to enhance portability of the HPC application deployment process across heterogeneous high-end machines. The SCVM approach to portable builds is based on the insertion of toolkit-interpretable directives into original application build scripts. Modifications resulting from these directives preserve the semantics of the original build instruction flow. The execution of the build script is controlled by our toolkit that intercepts build script commands in a manner transparent to the end-user. We have applied this approach to a scientific production code (Gamess-US) on the Cray-XT5 machine. The second facet, termed Unibus, aims to facilitate provisioning and aggregation of multifaceted resources from resource providers and end-users perspectives. To achieve that, Unibus proposes a Capability Model and mediators (resource drivers) to virtualize access to diverse resources, and soft and successive conditioning to enable automatic and user-transparent resource provisioning. A proof of concept implementation has demonstrated the viability of this approach on high end machines, grid systems and computing clouds.
LibVMI is an introspection library focused on reading and writing memory from virtual machines (VMs). For convenience, LibVMI also provides functions for accessing CPU registers, pausing and unpausing a VM, printing binary data, and more. LibVMI is designed to work across multiple virtualization platforms. LibVMI currently supports VMs running in either Xen or KVM. LibVMI also supports reading physical memory snapshots when saved as a file.
LibVMI is an introspection library focused on reading and writing memory from virtual machines (VMs). For convenience, LibVMI also provides functions for accessing CPU registers, pausing and unpausing a VM, printing binary data, and more. LibVMI is designed to work across multiple virtualization platforms. LibVMI currently supports VMs running in either Xen or KVM. LibVMI also supports reading physical memory snapshots when saved as a file.
Historically, high energy physics computing has been performed on large purpose-built computing systems. These began as single-site compute facilities, but have evolved into the distributed computing grids used today. Recently, there has been an exponential increase in the capacity and capability of commercial clouds. Cloud resources are highly virtualized and intended to be able to be flexibly deployed for a variety of computing tasks. There is a growing interest among the cloud providers to demonstrate the capability to perform large-scale scientific computing. In this paper, we discuss results from the CMS experiment using the Fermilab HEPCloud facility, which utilized both local Fermilab resources and virtual machines in the Amazon Web Services Elastic Compute Cloud. We discuss the planning, technical challenges, and lessons learned involved in performing physics workflows on a large-scale set of virtualized resources. Additionally, we will discuss the economics and operational efficiencies when executing workflows both in the cloud and on dedicated resources.
Cloud networks utilize virtual connections to connect virtual machines distributed across cloud sites. They are increasingly deployed due to flexible provisioning using software and cost-effectiveness in not requiring to build physical network infrastructure. However, their extensive virtualization makes it unclear how well the established practices of conventional networks translate to them. Here, we study throughput measurements over a Google Cloud network using a matching hardware emulated conventional network, which provide production and exploratory conditions, respectively. The measurements span connections representing local, cross-continental and around the Earth distances. We study the effects of parallel flows, congestion control algorithms and retransmissions on the network throughput profile expressed as a function of RTT. We compare the throughput profile of Google Cloud network with those of emulated network under various loss conditions, including those too disruptive or expensive in the former. Our analysis based on the concave-convex shape and utilization-concavity coefficients of throughput profiles indicates an overall agreement of performance between the two networks, thereby justifying the use of conventional network emulations to analyze cloud networks. In terms of practical use, our study establishes that BBR and BBRv2 alpha TCP achieve higher throughput compared to loss-based congestion control algorithms under most network configurations, especially, under losses at large RTT.
In this paper, we present a quantitative performance analysis of data analytics applications running on multi-core virtual machines. Such environments form the core of cloud computing. In addition, data analytics applications, such as Cassandra and Hadoop, are becoming increasingly popular on cloud computing platforms. This convergence necessitates a better understanding of the performance and cost implications of such hybrid systems. For example, the very rst step in hosting applications in virtualized environments, requires the user to con gure the number of virtual processors and the size of memory. To understand performance implications of this step, we benchmarked three Yahoo Cloud Serving Benchmark (YCSB) workloads in a virtualized multi-core environment. Our measurements indicate that the performance of Cassandra for YCSB workloads does not heavily depend on the processing capacity of a system, while the size of the data set is critical to performance relative to allocated memory. We also identi ed a strong relationship between the running time of workloads and various hardware events (last level cache loads, misses, and CPU migrations). From this analysis, we provide several suggestions to improve the performance of data analytics applications running on cloud computing environments.
We present Adrastea, an efficient FPGA design environment for developing scientific machine learning applications. FPGA development is challenging, from deployment, proper toolchain setup, programming methods, interfacing FPGA kernels, and more importantly, the need to explore design space choices to get the best performance and area usage from the FPGA kernel design. Adrastea provides an automated and scalable design flow to parameterize, implement, and optimize complex FPGA kernels and associated interfaces. We show how virtualization of the development environment via virtual machines is leveraged to simplify the setup of the FPGA toolchain while deploying the FPGA boards and while scaling up the automated design space exploration to leverage multiple machines concurrently. Adrastea provides an automated build and test environment of FPGA kernels. By exposing design space hyper-parameters, Adrastea can automatically search the design space in parallel to optimize the FPGA design for a given metric, usually performance or area. Adrastea simplifies the task of interfacing with the FPGA kernels with a simplified interface API. To demonstrate the capabilities of Adrastea, we implement a complex random forest machine learning kernel with 10,000 input features while achieving extremely low computing latency without loss of prediction accuracy, which is required by a scientific edge application at SNS. We also demonstrate Adrastea using an FFT kernel and show that for both applications Adrastea is able to systematically and efficiently evaluate different design options, which reduced the time and effort required to develop the kernel from months of manual work to days of automatic builds.