Engineering Papers⌕ Search

Engineering topics

Settlemyer, Bradley

Publications and source records attributed to Settlemyer, Bradley.

ZFS Interface For Accelerators

As data volume increases in HPC centers, storing data in reasonable amounts of space and time while maintaining data integrity and recoverability becomes ever more difficult. ZFS is a filesystem that provides these features, among many others, that make it attractive for usage in HPC environments. It has the ability to compress data, compute checksums, and compute redundancy codes within the same runtime, rather than compute each separately, without knowledge of the underlying structure of the filesystem. However, measurements have shown that compression on CPUs can be incredibly inefficient and significantly reduce the performance of ZFS. The ZFS Interface for Accelerators (Z.I.A.) was developed to provide an interface to route data to accelerators while being processed by ZFS, so that the ZFS infrastructure and features are maintained, while allowing for ZFS administrators to provide faster implementations of compression, or other features, to their instance of ZFS.

Lee, Jason↗

Data Processing Unit Services Module

The Data Processing Services Module (DPUSM) provides the ability to perform pluggable compression, erasure coding, checksuming and other important file system operations within the Linux kernel. The pluggable provider interface allows for the use of hardware acceleration of those services. In-kernel file systems are then able to use these functions to use these accelerators to perform operations that are normally run on the processor, resulting in improved file system performance. Third parties will register "providers" with the DPUSM to communicate with their respective accelerators. Providers will implement functions with DPUSM API signatures so that the DPUSM can translate the data inputted by users of the DPUSM into data that providers recognize.

Lee, Jason↗

It’s Time to Talk About HPC Storage: Perspectives on the Past and Future

High-performance computing (HPC) storage systems are a key component of the success of HPC to date. Recently, we have seen major developments in storage-related technologies, as well as changes to how HPC platforms are used, especially in relation to artificial intelligence and experimental data analysis workloads. Additionally, these developments merit a revisit of HPC storage system architectural designs. In this article, we discuss the drivers, identify key challenges to status quo posed by these developments, and discuss directions future research might take to unlock the potential of new technologies for the breadth of HPC applications.

97 MATHEMATICS AND COMPUTING↗

Mochi: Composing Data Services for High-Performance Computing Environments

Technology enhancements and the growing breadth of application workflows running on high-performance computing (HPC) platforms drive the development of new data services that provide high performance on these new platforms, provide capable and productive interfaces and abstractions for a variety of applications, and are readily adapted when new technologies are deployed. The Mochi framework enables composition of specialized distributed data services from a collection of connectable modules and subservices. Rather than forcing all applications to use a one-size-fits-all data staging and I/O software configuration, Mochi allows each application to use a data service specialized to its needs and access patterns. This paper introduces the Mochi framework and methodology. The Mochi core components and microservices are described. Examples of the application of the Mochi methodology to the development of four specialized services are detailed. Finally, a performance evaluation of a Mochi core component, a Mochi microservice, and a composed service providing an object model is performed. The paper concludes by positioning Mochi relative to related work in the HPC space and indicating directions for future work.

97 MATHEMATICS AND COMPUTING↗

A Software Defined Storage Approach to Exascale Storage Services (Final Technical Report)

The Mochi project began in 2015 in recognition of a set of changes that were set to alter the landscape of storage and I/O for DOE science. On the one hand, an influx of new classes of applications including an initial set of data intensive and machine learning applications were operating with new types of data and in new ways, as compared to traditional computational science simulation codes. Simultaneously, new memory, storage, and networking technologies were driving a need to adapt software designs to a rapidly changing design space. The goal of the three-year Mochi project was to develop and demonstrate an approach for composing storage services which provides new levels of functionality, performance, and reliability for science applications at extreme scale. The team, including a mix of Laboratory, university, and industry collaborators, focused on systems and applications of interest to DOE Office of Science, tracking developments at NERSC and the LCFs as well as application trends within ECP and SciDAC. Technical work focused on three themes: composition (understand the building blocks needed to provide storage services that meet DOE mission needs and how these components are flexibly composed), application interfacing (what service compositions are most effective for common DOE application use cases), and consistency and synchronization (taking advantage of knowledge about, and from, the system and scientific applications to provide better service than would otherwise be possible). The project was highly successful, producing 13 publications and tens of thousands of lines of open source software that is now being used by multiple ECP teams, HPC software vendors, and international collaborators. The uptake of Mochi technologies speaks to both the value of the service composition model and the quality of architecture and software developed by the team. Through continued funding under ASCR and ECP, the Mochi team and its technologies are on track to further contribute to DOE science successes into the exascale timeframe.

97 MATHEMATICS AND COMPUTING↗

Storage Systems and Input/Output: Organizing, Storing, and Accessing Data for Scientific Discovery. Report for the DOE ASCR Workshop on Storage Systems and I/O. [Summary Brief]

In September, 2018, the Department of Energy, Office of Science, Advanced Scientific Computing Research Program convened a workshop to identify key challenges and define research directions that will advance the field of storage systems and I/O over the next 5-7 years.The workshop concluded that addressing these combined challenges and opportunities requires tools and techniques that greatly extend traditional approaches and require new research directions. Key research opportunities were identified. This summary brief highlights three areas: The Impact of Storage and I/O, Research Opportunities, and, A new and productive Storage and I/O Environment. The full report from the workshop can be viewed at DOI: 10.2172/1491994.

97 MATHEMATICS AND COMPUTING↗

Storage Systems and Input/Output: Organizing, Storing, and Accessing Data for Scientific Discovery. Report for the DOE ASCR Workshop on Storage Systems and I/O. [Full Workshop Report]

In September, 2018, the Department of Energy, Office of Science, Advanced Scientific Computing Research Program convened a workshop to identify key challenges and define research directions that will advance the field of storage systems and I/O over the next 5–7 years. The workshop concluded that addressing these combined challenges and opportunities requires tools and techniques that greatly extend traditional approaches and require new research directions. Key research opportunities were identified.

97 MATHEMATICS AND COMPUTING↗

Regression-Based Analytics for Response Dynamics of SDN Solutions and Components

We consider Software-Defined Network (SDN) solutions that dynamically setup and teardown connections, and also their response dynamics of connectivity and TCP throughput. There is a wide variety of choices for controllers, switches, andsoftware modules, and objective analytics of measurements are needed to compare the response dynamics of solutions based on their combinations. Throughput and ping measurements during connection changes contain significant statistical variations due to the complexities of controllers, northbound scripts, and network devices. We propose analytical methods to estimate response regression functions, which provide a rigorous and objective comparison of different SDN solutions and their components, namely, controllers, switches, and OpenFlow versions. We apply this approach in two scenarios: (a) switching of long-haul connections emulated in hardware by ANUE devices over a testbed consisting of HP and Cisco switches, and (b) multi-site science infrastructure emulated using VMs and Mininet. The results provide useful practical insights including: (i) the dpctl method responds seconds faster than the OpenDayLight (ODL) method on average under OpenFlow 1.0, but performs similar to ODL under OpenFlow 1.3, and (ii) the ping response of dpctl is about 0.6 sec faster than ODL for the second scenario.

97 MATHEMATICS AND COMPUTING↗

Virtual Environment for Testing Software-Defined Networking Solutions for Scientific Workflows

Recent developments in software-defined infrastructures promise that scientific workflows utilizing supercomputers, instruments, and storage systems will be dynamically composed and orchestrated using software at unprecedented speed and scale in the near future. Testing of the underlying networking software, particularly during initial exploratory stages, remains a challenge due to potential disruptions, and resource allocation and coordination needed over the multi-domain physical infrastructure. To overcome these challenges, we develop the Virtual Science Network Environment (VSNE) that emulates the multi-site host, storage, and network infrastructure using Virtual Machines (VMs), wherein the production and nascent software can be tested. Within each VM, which represents a site, the hosts and local-area networks are emulated using Mininet, and the Software-Defined Network (SDN) controllers and service daemon codes are natively run to support dynamic provisioning of network connections. Additionally, Lustre filesystem support at the sites and an emulation of the long-haul network using Mininet, are provided using separate VMs. As case studies, we describe Lustre file transfers using XDD, Red5 streaming service demonstration, and an emulated experiment with remote monitoring and steering modules, all supported over dynamically configured connections using SDN controllers.

97 MATHEMATICS AND COMPUTING↗

So ware-Defined Network Solutions for Science Scenarios: Performance Testing Framework and Measurements

High-performance scientific work flows utilize supercomputers, scientific instruments, and large storage systems. Their executions require fast setup of a small number of dedicated network connections across the geographically distributed facility sites. We present Software-Defined Network (SDN) solutions consisting of site daemons that use dpctl, Floodlight, ONOS, or OpenDaylight controllers to set up these connections. The development of these SDN solutions could be quite disruptive to the infrastructure, while requiring a close coordination among multiple sites; in addition, the large number of possible controller and device combinations to investigate could make the infrastructure unavailable to regular users for extended periods of time. In response, we develop a Virtual Science Network Environment (VSNE) using virtual machines, Mininet, and custom scripts that support the development, testing, and evaluation of SDN solutions, without the constraints and expenses of multi-site physical infrastructures; furthermore, the chosen solutions can be directly transferred to production deployments. By complementing VSNE with a physical testbed, we conduct targeted performance tests of various SDN solutions to help choose the best candidates. In addition, we propose a switching response method to assess the setup times and throughput performances of different SDN solutions, and present experimental results that show their advantages and limitations.

Settlemyer, Bradley↗

Experimental Analysis of File Transfer Rates over Wide-Area Dedicated Connections

File transfers over dedicated connections, supported by large parallel file systems, have become increasingly important in high-performance computing and big data workflows. It remains a challenge to achieve peak rates for such transfers due to the complexities of file I/O, host, and network transport subsystems, and equally importantly, their interactions. We present extensive measurements of disk-to-disk file transfers using Lustre and XFS file systems mounted on multi-core servers over a suite of 10 Gbps emulated connections with 0-366 ms round trip times. Our results indicate that large buffer sizes and many parallel flows do not always guarantee high transfer rates. Furthermore, large variations in the measured rates necessitate repeated measurements to ensure confidence in inferences based on them. We propose a new method to efficiently identify the optimal joint file I/O and network transport parameters using a small number of measurements. We show that for XFS and Lustre with direct I/O, this method identifies configurations achieving 97% of the peak transfer rate while probing only 12% of the parameter space.

Rao, Nageswara S.↗

High-Performance data flows using analytical models and measurements

The combination of analytical models and measurements provide practical configurations and parameters to achieve high data transport rates: (a) buffer sizes and number of parallel streams for improved memory and file transfer rates, (b) Hamilton and Scalable TCP congestion control modules for memory transfers in place of default CUBIC, and (c) direct IO mode for Lustre file systems for wide-area transfers. Conventional parameter selection using full sweeps is impractical in many cases since it takes months. By exploiting the unimodality of throughput profiles, we developed the d-w method that significantly reduces the number of measurements needed for parameter identification. This heuristic method was effective in practice in reducing the measurements by about 90% for Lustre and XFS file transfers.

Rao, Nageswara S↗

Measurements of file transfer rates over dedicated long-haul connections

Wide-area file transfers are an integral part of several High-Performance Computing (HPC) scenarios. Dedicated network connections with high capacity, low loss rate and low competing traffic, are increasingly being provisioned over current HPC infrastructures to support such transfers. To gain insights into these file transfers, we collected transfer rate measurements for Lustre and xfs file systems between dedicated multi-core servers over emulated 10 Gbps connections with round trip times (rtt) in 0-366 ms range. Memory transfer throughput over these connections is measured using iperf, and file IO throughput on host systems is measured using xddprof. We consider two file system configurations: Lustre over IB network and xfs over SSD connected to PCI bus. Files are transferred using xdd across these connections, and the transfer rates are measured, which indicate the need to jointly optimize the connection and host file IO parameters to achieve peak transfer rates. In particular, these measurements indicate that (i) peak file transfer rate is lower than peak connection and host IO throughput, in some cases by as much as 50% or lower, (ii) xdd request sizes that achieve peak throughput for host file IO do not necessarily lead to peak file transfer rates, and (iii) parallelism in host IO and TCP transport does not always improve the file transfer rates.

Rao, Nageswara S↗

SSD-optimized workload placement with adaptive learning and classification in HPC environments

In recent years, non-volatile memory devices such as SSD drives have emerged as a viable storage solution due to their increasing capacity and decreasing cost. Due to the unique capability and capacity requirements in large scale HPC (High Performance Computing) storage environment, a hybrid configuration (SSD and HDD) may represent one of the most available and balanced solutions considering the cost and performance. Under this setting, effective data placement as well as movement with controlled overhead become a pressing challenge. In this paper, we propose an integrated object placement and movement framework and adaptive learning algorithms to address these issues. Specifically, we present a method that shuffle data objects across storage tiers to optimize the data access performance. The method also integrates an adaptive learning algorithm where realtime classification is employed to predict the popularity of data object accesses, so that they can be placed on, or migrate between SSD or HDD drives in the most efficient manner. We discuss preliminary results based on this approach using a simulator we developed to show that the proposed methods can dynamically adapt storage placements and access pattern as workloads evolve to achieve the best system level performance such as throughput.

Wan, Lipeng↗