Search NASA⌕ Search

SEARCH · Search NASA

Results for “Workload”

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.

At least 1,495 records · Page 83

Small File Aggregation with PLFS [Slides]

Today’s computational science demands have resulted in ever larger parallel computers, and storage systems have grown to match these demands. Parallel file systems used in this environment are increasingly specialized to extract the highest possible performance for large I/O operations, at the expense of other potential workloads. While some applications have adapted to I/O best practices and can obtain good performance on these systems, the natural I/O patterns of many applications result in the generation of a huge number of small files, the creation of which is poorly served by current parallel file systems at very large scale. This paper describes a new technique for optimizing small file access in parallel file systems for these very large scale systems. The idea is to use a virtual parallel log-structure file system on the compute nodes in order to aggregate large numbers of small files in compute node memory and then stream their data sequentially to a much smaller number of physical files on an underlying parallel file system. The technique is implemented and evaluated using PLFS as the aggregating middleware. We evaluate our system with micro-benchmarks on a local OSX filesystem and with an MPI extension of the standard Postmark to provide results at scale on both Lustre and PanFS parallel filesystems. We observe as much as a 33x improvement in small file create rates on a single host, and 30x improvement in small file write rates, compared to a baseline Lustre configuration on a leadership computing platform using 16,384 cores and achieve an unprecedented create rate of 200 million files per second.

97 MATHEMATICS AND COMPUTING↗

The Magellan Final Report on Cloud Computing

The goal of Magellan, a project funded through the U.S. Department of Energy (DOE) Office of Advanced Scientific Computing Research (ASCR), was to investigate the potential role of cloud computing in addressing the computing needs for the DOE Office of Science (SC), particularly related to serving the needs of mid- range computing and future data-intensive computing workloads. A set of research questions was formed to probe various aspects of cloud computing from performance, usability, and cost. To address these questions, a distributed testbed infrastructure was deployed at the Argonne Leadership Computing Facility (ALCF) and the National Energy Research Scientific Computing Center (NERSC). The testbed was designed to be flexible and capable enough to explore a variety of computing models and hardware design points in order to understand the impact for various scientific applications. During the project, the testbed also served as a valuable resource to application scientists. Applications from a diverse set of projects such as MG-RAST (a metagenomics analysis server), the Joint Genome Institute, the STAR experiment at the Relativistic Heavy Ion Collider, and the Laser Interferometer Gravitational Wave Observatory (LIGO), were used by the Magellan project for benchmarking within the cloud, but the project teams were also able to accomplish important production science utilizing the Magellan cloud resources.

97 MATHEMATICS AND COMPUTING↗

Automatic Transformation of MPI Programs to Asynchronous, Graph-Driven Form

The goals of this project are to develop new, scalable, high-fidelity algorithms for atomic-level simulations and program transformations that automatically restructure existing applications, enabling them to scale forward to Petascale systems and beyond. The techniques enable legacy MPI application code to exploit greater parallelism though increased latency hiding and improved workload assignment. The techniques were successfully demonstrated on high-end scalable systems located at DOE laboratories. Besides the automatic MPI program transformations efforts, the project also developed several new scalable algorithms for ab-initio molecular dynamics, including new massively parallel algorithms for hybrid DFT and new parallel in time algorithms for molecular dynamics and ab-initio molecular dynamics. These algorithms were shown to scale to very large number of cores, and they were designed to work in the latency hiding framework developed in this project. The effectiveness of the developments was enhanced by the direct application to real grand challenge simulation problems covering a wide range of technologically important applications, time scales and accuracies. These included the simulation of the electronic structure of mineral/fluid interfaces, the very accurate simulation of chemical reactions in microsolvated environments, and the simulation of chemical behavior in very large enzyme reactions.

97 MATHEMATICS AND COMPUTING↗

Virtual tool mark generation for efficient striation analysis in forensic science

In 2009, a National Academy of Sciences report called for investigation into the scienti c basis behind tool mark comparisons (National Academy of Sciences, 2009). Answering this call, Chumbley et al. (2010) attempted to prove or disprove the hypothesis that tool marks are unique to a single tool. They developed a statistical algorithm that could, in most cases, discern matching and non-matching tool marks made at di erent angles by sequentially numbered screwdriver tips. Moreover, in the cases where the algorithm misinterpreted a pair of marks, an experienced forensics examiner could discern the correct outcome. While this research served to con rm the basic assumptions behind tool mark analysis, it also suggested that statistical analysis software could help to reduce the examiner's workload. This led to a new tool mark analysis approach, introduced in this thesis, that relies on 3D scans of screwdriver tip and marked plate surfaces at the micrometer scale from an optical microscope. These scans are carefully cleaned to remove noise from the data acquisition process and assigned a coordinate system that mathematically de nes angles and twists in a natural way. The marking process is then simulated by using a 3D graphics software package to impart rotations to the tip and take the projection of the tip's geometry in the direction of tool travel. The edge of this projection, retrieved from the 3D graphics software, becomes a virtual tool mark. Using this method, virtual marks are made at increments of 5 and compared to a scan of the evidence mark. The previously developed statistical package from Chumbley et al. (2010) performs the comparison, comparing the similarity of the geometry of both marks to the similarity that would occur due to random chance. The resulting statistical measure of the likelihood of the match informs the examiner of the angle of the best matching virtual mark, allowing the examiner to focus his/her mark analysis on a smaller range of angles. Preliminary results are quite promising. In a study with both sides of 6 screwdriver tips and 34 corresponding marks, the method distinguished known matches from known non-matches with zero false positive matches and only two matches mistaken for non-matches. For matches, it could predict the correct marking angle within 5-10 . Moreover, on a standard desktop computer, the virtual marking software is capable of cleaning 3D tip and plate scans in minutes and producing a virtual mark and comparing it to a real mark in seconds. These results support several of the professional conclusions of the tool mark analysis com- munity, including the idea that marks produced by the same tool only match if they are made at similar angles. The method also displays the potential to automate part of the comparison process, freeing the examiner to focus on other tasks, which is important in busy, backlogged crime labs. Finally, the method o ers the unique chance to directly link an evidence mark to the tool that produced it while reducing potential damage to the evidence.

42 ENGINEERING↗

Application of Subspace Detectors to 2012 Sumatera Earthquake Sequence

In this report the aftershock sequence of the 2012 Sumatra earthquake is used to study the performance of subspace detectors (Harris, 2006) to detect and classify the events. The ground truth used for this study is 1222 aftershock solutions from the LLNL database drawn mostly from the Reviewed Event Bulletin (REB). Subspace detectors are built using several spawning strategies with waveforms recorded by the Makanchi Array (MKAR). Because we desire the subspace detectors to have a very low false alarm rate, the subspace templates are required to be about 50 seconds long (most of the P-bundle). For the same reason, the power detections used to create the subspace templates are restricted to have high SNR. Because of these restrictions, none of the subspace detectors are able to reproduce the entire ground truth catalog. They do have a very low false alarm rate, however, and so can reliably be used as single-array detection classifiers. Because of their reliability, in a suitably designed system, these detectors could be used to segregate detections from regions of high seismicity such as aftershock sequences, and allow processing of those detections in isolation from routine seismicity. It may also be possible to use these detectors to process groups of detections simultaneously. This leads to the concept of “analyst workload reduction factor”; a measure of the reduction in effort achieved by processing N detections in M groups instead of individually. If M is much smaller than N, a significant reduction factor is achieved.

58 GEOSCIENCES↗

KONGMING: Performance Prediction in the Cloud via Multidimensional Interference Surrogates

As more and more applications are deployed in the cloud, it is important for both the user and the operator of the cloud that the resources of the cloud are utilized efficiently. Virtualization and workload consolidation techniques are pervasively applied in the cloud to increase resource utilization while providing isolated execution environments for different users. While virtualization hides the architectural details of the underlying hardware, it can also increase the variability in application execution times due to heterogeneity in available hardware, and interference from other applications sharing the same hardware resources. This reduces both the productivity of cloud platforms and limits the degree to which software colocation can be used to increase its efficiency.

97 MATHEMATICS AND COMPUTING↗

A Computuerized Operator Support System Prototype

A report was published by the Idaho National Laboratory in September of 2012, entitled Design to Achieve Fault Tolerance and Resilience, which described the benefits of automating operator actions for transients. The report identified situations in which providing additional automation in lieu of operator actions would be advantageous. It recognized that managing certain plant upsets is sometimes limited by the operator’s ability to quickly diagnose the fault and to take the needed actions in the time available. Undoubtedly, technology is underutilized in the nuclear power industry for operator assistance during plant faults and operating transients. In contrast, other industry sectors have amply demonstrated that various forms of operator advisory systems can enhance operator performance while maintaining the role and responsibility of the operator as the independent and ultimate decision-maker. A computerized operator support system (COSS) is proposed for use in nuclear power plants to assist control room operators in addressing time-critical plant upsets. A COSS is a collection of technologies to assist operators in monitoring overall plant performance and making timely, informed decisions on appropriate control actions for the projected plant condition. The COSS does not supplant the role of the operator, but rather provides rapid assessments, computations, and recommendations to reduce workload and augment operator judgment and decision-making during fast-moving, complex events. This project proposes a general model for a control room COSS that addresses a sequence of general tasks required to manage any plant upset: detection, validation, diagnosis, recommendation, monitoring, and recovery. The model serves as a framework for assembling a set of technologies that can be interrelated to assist with each of these tasks. A prototype COSS has been developed in order to demonstrate the concept and provide a test bed for further research. The prototype is based on four underlying elements consisting of a digital alarm system, computer-based procedures, PI&D system representations, and a recommender module for mitigation actions. At this point, the prototype simulates an interface to a sensor validation module and a fault diagnosis module. These two modules will be fully integrated in the next version of the prototype. The initial version of the prototype is now operational at the Idaho National Laboratory using the U.S. Department of Energy’s Light Water Reactor Sustainability (LWRS) Human Systems Simulation Laboratory (HSSL). The HSSL is a full-scope, full-scale glass top simulator capable of simulating existing and future nuclear power plant main control rooms. The COSS is interfaced to the Generic Pressurized Water Reactor (gPWR) simulator with industry-typical control board layouts. The glass top panels display realistic images of the control boards that can be operated by touch gestures. A section of the simulated control board was dedicated to the COSS human-system interface (HSI), which resulted in a seamless integration of the COSS into the normal control room environment.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

A Computuerized Operator Support System Prototype

A report was published by the Idaho National Laboratory in September of 2012, entitled Design to Achieve Fault Tolerance and Resilience, which described the benefits of automating operator actions for transients. The report identified situations in which providing additional automation in lieu of operator actions would be advantageous. It recognized that managing certain plant upsets is sometimes limited by the operator’s ability to quickly diagnose the fault and to take the needed actions in the time available. Undoubtedly, technology is underutilized in the nuclear power industry for operator assistance during plant faults and operating transients. In contrast, other industry sectors have amply demonstrated that various forms of operator advisory systems can enhance operator performance while maintaining the role and responsibility of the operator as the independent and ultimate decision-maker. A computerized operator support system (COSS) is proposed for use in nuclear power plants to assist control room operators in addressing time-critical plant upsets. A COSS is a collection of technologies to assist operators in monitoring overall plant performance and making timely, informed decisions on appropriate control actions for the projected plant condition. The COSS does not supplant the role of the operator, but rather provides rapid assessments, computations, and recommendations to reduce workload and augment operator judgment and decision-making during fast-moving, complex events. This project proposes a general model for a control room COSS that addresses a sequence of general tasks required to manage any plant upset: detection, validation, diagnosis, recommendation, monitoring, and recovery. The model serves as a framework for assembling a set of technologies that can be interrelated to assist with each of these tasks. A prototype COSS has been developed in order to demonstrate the concept and provide a test bed for further research. The prototype is based on four underlying elements consisting of a digital alarm system, computer-based procedures, PI&D system representations, and a recommender module for mitigation actions. At this point, the prototype simulates an interface to a sensor validation module and a fault diagnosis module. These two modules will be fully integrated in the next version of the prototype. The initial version of the prototype is now operational at the Idaho National Laboratory using the U.S. Department of Energy’s Light Water Reactor Sustainability (LWRS) Human Systems Simulation Laboratory (HSSL). The HSSL is a full-scope, full-scale glass top simulator capable of simulating existing and future nuclear power plant main control rooms. The COSS is interfaced to the Generic Pressurized Water Reactor (gPWR) simulator with industry-typical control board layouts. The glass top panels display realistic images of the control boards that can be operated by touch gestures. A section of the simulated control board was dedicated to the COSS human-system interface (HSI), which resulted in a seamless integration of the COSS into the normal control room environment.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

RBAC Driven Least Privilege Architecture For Control Systems

The concept of role based access control (RBAC) within the IT environment has been studied by researchers and was supported by NIST (circa 1992). This earlier work highlighted the benefits of RBAC which include reduced administrative workload and policies which are easier to analyze and apply. The goals of this research were to expand the application of RBAC in the following ways. Apply RBAC to the control systems environment: The typical RBAC model within the IT environment is used to control a user’s access to files. Within the control system environment files are replaced with measurement (e.g., temperature) and control (e.g. valve) points organized as a hierarchy of control assets (e.g. a boiler, compressor, refinery unit). Control points have parameters (e.g., high alarm limit, set point, etc.) associated with them. The RBAC model is extended to support access to points and their parameters based upon roles while at the same time allowing permissions for the points to be defined at the asset level or point level directly. In addition, centralized policy administration with distributed access enforcement mechanisms was developed to support the distributed architecture of distributed control systems and SCADA; Extend the RBAC model to include access control for software and devices: The established RBAC approach is to assign users to roles. This work extends that notion by first breaking the control system down into three layers 1) users, 2) software and 3) devices. An RBAC model is then created for each of these three layers. The result is that RBAC can be used to define machine-to-machine policy enforced via the IP security (IPsec) protocol. This highlights the potential to use RBAC for machine-to-machine connectivity within the internet of things; and Enable dynamic policy based upon the operating mode of the system: The IT environment is generally static with respect to policy. However, large cyber physical systems such as industrial controls have various operating modes (start-up, normal operation, emergency, shut-down and maintenance are typical). The policy enforcement architecture must be able to support changes in access permissions as the mode of the control system changes. For example an operator’s role may not allow the operator to shut down a pump during “normal operation” but that same operator role may be given permission to shut down the pump if the refinery transitions to “emergency” mode. The effectiveness of the approach was validated by applying it to the Experion Process Knowledge System. This is a large commercial industrial control system often used to control oil refineries and other assets within the oil and gas sector. As a by-product, other industries using Experion (Pharmaceuticals, Specialty Chemicals, etc.) also benefit from increased security. Policies representative of those that would be used within an oil refinery were created and validated against the RBAC model as implemented in the underlying SQL database. The administration of policy is simplified which in turn makes it practical for security administrators to specify policies which enforce least privilege. The result is a qualitative reduction in risk. The benefits of the enhanced RBAC model are clear and as a result. Honeywell is incorporating portions of the RBAC research into the 2014 release of Experion.

97 MATHEMATICS AND COMPUTING↗

MPI-Bash: Parallel scripting right from the Bourne-Again Shell (Bash)

Parallel, scientific applications running on massively parallel supercomputers commonly produce large numbers of large data files. While parallel filesystems improve the performance of file generation, postprocessing activities such as archiving and compressing the data are often run sequentially (and therefore slowly), squandering the supercomputer's vast performance. Consequently, data that take hours to generate may take days to postprocess. Writing and debugging parallel postprocessing programs is often too time-consuming to be worthwhile, especially for one-off postprocessing operations. What the community needs is a simple mechanism for cobbling together scripts that distribute postprocessing work across a supercomputer's many nodes. MPI-Bash attempts to supply that mechanism by augmenting the GNU project's command shell, Bash, with support for the Message Passing Interface (MPI) and, if available, the Libcircle library for work distribution and load balancing. MPI-Bash makes it easy to parallelize bash scripts that run a set of Linux commands independently over a large number of input files. Because MPI-Bash includes various MPI functions for data transfer and synchronization, it is not limited to embarrassingly parallel workloads but can incorporate phased operations (i.e., all workers must finish operation X before any worker is allowed to begin operation Y).

97 MATHEMATICS AND COMPUTING↗

An Advanced Network and distributed Storage Laboratory (ANDSL) for Data Intensive Science (Final Report)

The original intent of this project was to build and operate an Advanced Network and Distributed Storage Laboratory (ANDSL) for Data Intensive Science that will prepare the Open Science Grid (OSG) community for a new generation of wide area communication capabilities operating at a 100Gb rate. The plan was to engage a team of four dedicated scientists that will design, develop and operate a state of the art laboratory that will be fully integrated into the organizational and operational infrastructure of the OSG. The ANDSL was planned to include host-based hardware and software required to profile, emulate and study real-life distributed storage workloads at rates of 100 Gbps. The laboratory was designed to utilize the Department Of Energy’s Advanced Network Initiative (ANI) 100GbTestbed as connections to the fabric become available so that OSG stakeholders can use it to profile real-life data intensive science applications as well as experiment with advanced storage and data management technologies. The goal of the original program of work was to close the adoption gap, decreasing the time between availability of production 100Gbps links and the end-to-end use of such a network to enable scientific discovery.

100Gb networks↗

Light Water Reactor Sustainability Program: Digital Technology Business Case Methodology Guide

The Department of Energy’s (DOE’s) Light Water Reactor Sustainability Program aims to develop and deploy technologies that will make the existing U.S. nuclear fleet more efficient and competitive. The program has developed a standard methodology for determining the impact of new technologies in order to assist nuclear power plant (NPP) operators in building sound business cases. The Advanced Instrumentation, Information, and Control (II&C) Systems Technologies Pathway is part of the DOE’s Light Water Reactor Sustainability (LWRS) Program. It conducts targeted research and development (R&D) to address aging and reliability concerns with the legacy instrumentation and control and related information systems of the U.S. operating light water reactor (LWR) fleet. This work involves two major goals: (1) to ensure that legacy analog II&C systems are not life-limiting issues for the LWR fleet and (2) to implement digital II&C technology in a manner that enables broad innovation and business improvement in the NPP operating model. Resolving long-term operational concerns with the II&C systems contributes to the long-term sustainability of the LWR fleet, which is vital to the nation’s energy and environmental security. The II&C Pathway is conducting a series of pilot projects that enable the development and deployment of new II&C technologies in existing nuclear plants. Through the LWRS program, individual utilities and plants are able to participate in these projects or otherwise leverage the results of projects conducted at demonstration plants. Performance advantages of the new pilot project technologies are widely acknowledged, but it has proven difficult for utilities to derive business cases for justifying investment in these new capabilities. Lack of a business case is often cited by utilities as a barrier to pursuing wide-scale application of digital technologies to nuclear plant work activities. The decision to move forward with funding usually hinges on demonstrating actual cost reductions that can be credited to budgets and thereby truly reduce O&M or capital costs. Technology enhancements, while enhancing work methods and making work more efficient, often fail to eliminate workload such that it changes overall staffing and material cost requirements. It is critical to demonstrate cost reductions or impacts on non-cost performance objectives in order for the business case to justify investment by nuclear operators. The Business Case Methodology (BCM) addresses the “benefit” side of the analysis—as opposed to the cost side—and how the organization evaluates discretionary projects (net present value (NPV), accounting effects of taxes, discount rates, etc.). The cost and analysis side is not particularly difficult for the organization and can usually be determined with a fair amount of precision (not withstanding implementation project cost overruns). It is in determining the "benefits" side of the analysis that utilities have more difficulty in technology projects and that is the focus of this methodology.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Active Storage with Analytics Capabilities and I/O Runtime System for Petascale Systems

Computational scientists must understand results from experimental, observational and computational simulation generated data to gain insights and perform knowledge discovery. As systems approach the petascale range, problems that were unimaginable a few years ago are within reach. With the increasing volume and complexity of data produced by ultra-scale simulations and high-throughput experiments, understanding the science is largely hampered by the lack of comprehensive I/O, storage, acceleration of data manipulation, analysis, and mining tools. Scientists require techniques, tools and infrastructure to facilitate better understanding of their data, in particular the ability to effectively perform complex data analysis, statistical analysis and knowledge discovery. The goal of this work is to enable more effective analysis of scientific datasets through the integration of enhancements in the I/O stack, from active storage support at the file system layer to MPI-IO and high-level I/O library layers. We propose to provide software components to accelerate data analytics, mining, I/O, and knowledge discovery for large-scale scientific applications, thereby increasing productivity of both scientists and the systems. Our approaches include 1) design the interfaces in high-level I/O libraries, such as parallel netCDF, for applications to activate data mining operations at the lower I/O layers; 2) Enhance MPI-IO runtime systems to incorporate the functionality developed as a part of the runtime system design; 3) Develop parallel data mining programs as part of runtime library for server-side file system in PVFS file system; and 4) Prototype an active storage cluster, which will utilize multicore CPUs, GPUs, and FPGAs to carry out the data mining workload.

97 MATHEMATICS AND COMPUTING↗

Load Balancing Scientific Applications

The largest supercomputers have millions of independent processors, and concurrency levels are rapidly increasing. For ideal efficiency, developers of the simulations that run on these machines must ensure that computational work is evenly balanced among processors. Assigning work evenly is challenging because many large modern parallel codes simulate behavior of physical systems that evolve over time, and their workloads change over time. Furthermore, the cost of imbalanced load increases with scale because most large-scale scientific simulations today use a Single Program Multiple Data (SPMD) parallel programming model, and an increasing number of processors will wait for the slowest one at the synchronization points. To address load imbalance, many large-scale parallel applications use dynamic load balance algorithms to redistribute work evenly. The research objective of this dissertation is to develop methods to decide when and how to load balance the application, and to balance it effectively and affordably. We measure and evaluate the computational load of the application, and develop strategies to decide when and how to correct the imbalance. Depending on the simulation, a fast, local load balance algorithm may be suitable, or a more sophisticated and expensive algorithm may be required. We developed a model for comparison of load balance algorithms for a specific state of the simulation that enables the selection of a balancing algorithm that will minimize overall runtime.

97 MATHEMATICS AND COMPUTING↗

Scalability under a Power Bound using the GREMLINs Framework

With the move towards exascale, system and software developers will have to deal with issues of extreme parallelism. The system properties affected most by the increase in node and core count are the shared resources on node and across the system. The increase in parallelism leads to reduced memory and bandwidth when regarding individual cores. Since power is a limiting factor for supercomputers, and power is not fully utilized in current systems, overprovisioning compute resources is a viable approach to maximized power utilization. To maximize system performance in regard to these changing conditions, it is necessary to understand how resource restrictions impact performance and system behavior. For the purpose of understanding anticipated system properties the GREMLINs Framework was developed. The framework gives the opportunity to add power restrictions, hinder memory properties and introduce faults to study resilience, among others. These features give the opportunity to use current petascale technology to study problems system designers and software developers will have to face when moving towards exascale and beyond. This work describes the initial release of the GREMLINs Framework, developed for this work, and shows how it can be used to study the scaling behavior of proxy applications. These proxy applications represent a selection of HPC workloads important to the scientific community. The proxy applications studied are AMG2013, an algebraic multi-grid linear system solver, CoMD, a classical molecular dynamics proxy application and NEKBONE, an application that uses a high order spectral element method to solve the Navier-Stokes equations. The main interest of these studies lies in analysis regarding their power behavior at scale under a power bound. These findings show how the GREMLINs Framework can help systems and software designers to attain better application performance and can also be used as basis for CPU power balancing tools to use power more efficiently.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Review of Enabling Technologies to Facilitate Secure Compute Customization

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...

Aderholdt, Ferrol↗

Multi-Tenant Isolation via Reconfigurable Networks

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 at various security levels but in so doing are often enclaved at the highest security posture. This approach places significant restrictions on the users of the system even when processing data at a lower security level and exposes data at higher levels of confidentiality to a much broader population than otherwise necessary. The traditional approach of isolation, while effective in establishing security enclaves poses significant challenges for the use of shared infrastructure in HPC environments. This report details current state-of-the-art in reconfigurable network enclaving through Software Defined Networking (SDN) and Network Function Virtualization (NFV) and their applicability to secure enclaves in HPC environments. SDN and NFV methods are based on a solid foundation of system wide virtualization. The purpose of which is very straight forward, the system administrator can deploy networks that are more amenable to customer needs, and at the same time achieve increased scalability making it easier to increase overall capacity as needed without negatively affecting functionality. The network administration of both the server system and the virtual sub-systems is simplified allowing control of the infrastructure through well-defined APIs (Application Programming Interface). While SDN and NFV technologies offer significant promise in meeting these goals, they also provide the ability to address a significant component of the multi-tenant challenge in HPC environments, namely resource isolation. Traditional HPC systems are built upon scalable high-performance networking technologies designed to meet specific application requirements. Dynamic isolation of resources within these environments has remained difficult to achieve. SDN and NFV methodology provide us with relevant concepts and available open standards based APIs that isolate compute and storage resources within an otherwise common networking infrastructure. Additionally, the integration of the networking APIs within larger system frameworks such as OpenStack provide the tools necessary to establish isolated enclaves dynamically allowing the benefits of HPC while providing a controlled security structure surrounding these systems.

97 MATHEMATICS AND COMPUTING↗

Evaluating Storage Systems for Lustre

Storage systems are complex, including multiple subsystems and components. Sustained operations with top performance require all these subsystems and components working as expected. Having a detailed performance profile helps establishing a baseline. This baseline can be used for easier identification of possible future problems. A systematic bottom-to-top approach, starting with a detailed performance analysis of disks and moving up across layers and subsystems, provides a quantitative breakdown of each component's capabilities and bottlenecks. Coupling these low-level tests with Lustre-level evaluations will present a better understanding of performance expectations under different I/O workloads.

97 MATHEMATICS AND COMPUTING↗