Publications

Abstract not provided.

Large-scale, high-throughput computational science faces an accelerating convergence of software and hardware. Software container-based solutions have become common in cloud-based datacenter environments, and are considered promising tools for addressing heterogeneity and portability concerns. However, container solutions reflect a set of assumptions which complicate their adoption by developers and users of scientific workflow applications. Nor are containers a universal solution for deployment in high-performance computing (HPC) environments which have specialized and vertically integrated scheduling and runtime software stacks. In this paper, we present a container design and deployment approach which uses modular layering to ease the deployment of containers into existing HPC environments. This layered approach allows operating system integrations, support for different communication and performance monitoring libraries, and application code to be defined and interchanged in isolation. We describe in this paper the details of our approach, including specifics about container deployment and orchestration for different HPC scheduling systems. We also describe how this layering method can be used to build containers for two separate applications, each deployed on clusters with different batch schedulers, MPI networking support, and performance monitoring requirements. Our experience indicates that the layered approach is a viable strategy for building applications intended to provide similar behavior across widely varying deployment targets.

Remote Direct Memory Access (RDMA) capabilities have been provided by high-end networks for many years, but the network environments surrounding RDMA are evolving. RDMA performance has historically relied on using strict ordering guarantees to determine when data transfers complete, but modern adaptively-routed networks no longer provide those guarantees. RDMA also exposes low-level details about memory buffers: either all clients are required to coordinate access using a single shared buffer, or exclusive resources must be allocatable per-client for an unbounded amount of time. This makes RDMA unattractive for use in many-to-one communication models such as those found in public internet client-server situations.Remote Virtual Memory Access (RVMA) is a novel approach to data transfer which adapts and builds upon RDMA to provide better usability, resource management, and fault tolerance. RVMA provides a lightweight completion notification mechanism which addresses RDMA performance penalties imposed by adaptively-routed networks, enabling high-performance data transfer regardless of message ordering. RVMA also provides receiver-side resource management, abstracting away previously-exposed details from the sender-side and removing the RDMA requirement for exclusive/coordinated resources. RVMA requires only small hardware modifications from current designs, provides performance comparable or superior to traditional RDMA networks, and offers many new features.In this paper, we describe RVMA's receiver-managed resource approach and how it enables a variety of new data-transfer approaches on high-end networks. In particular, we demonstrate how an RVMA NIC could implement the first hardware-based fault tolerant RDMA-like solution. We present the design and validation of an RVMA simulation model in a popular simulation suite and use it to evaluate the advantages of RVMA at large scale. In addition to support for adaptive routing and easy programmability, RVMA can outperform RDMA on a 3D sweep application by 4.4X.

Persistent memory (PMEM) devices can achieve comparable performance to DRAM while providing significantly more capacity. This has made the technology compelling as an expansion to main memory. Rethinking PMEM as storage devices can offer a high performance buffering layer for HPC applications to temporarily, but safely store data. However, modern parallel I/O libraries, such as HDF5 and pNetCDF, are complicated and introduce significant software and metadata overheads when persisting data to these storage devices, wasting much of their potential. In this work, we explore the potential of PMEM as storage through pMEMCPY: a simple, lightweight, and portable I/O library for storing data in persistent memory. We demonstrate that our approach is up to 2x faster than other popular parallel I/O libraries under real workloads.

The FAIR principles of open science (Findable, Accessible, Interoperable, and Reusable) have had transformative effects on modern large-scale computational science. In particular, they have encouraged more open access to and use of data, an important consideration as collaboration among teams of researchers accelerates and the use of workflows by those teams to solve problems increases. How best to apply the FAIR principles to workflows themselves, and software more generally, is not yet well understood. We argue that the software engineering concept of technical debt management provides a useful guide for application of those principles to workflows, and in particular that it implies reusability should be considered as 'first among equals'. Moreover, our approach recognizes a continuum of reusability where we can make explicit and selectable the tradeoffs required in workflows for both their users and developers. To this end, we propose a new abstraction approach for reusable workflows, with demonstrations for both synthetic workloads and real-world computational biology workflows. Through application of novel systems and tools that are based on this abstraction, these experimental workflows are refactored to rightsize the granularity of workflow components to efficiently fill the gap between end-user simplicity and general customizability. Our work makes it easier to selectively reason about and automate the connections between trade-offs across user and developer concerns when exposing degrees of freedom for reuse. Additionally, by exposing fine-grained reusability abstractions we enable performance optimizations, as we demonstrate on both institutional-scale and leadership-class HPC resources.

Performance variation deriving from hardware and software sources is common in modern scientific and data-intensive computing systems, and synchronization in parallel and distributed programs often exacerbates their impacts at scale. The decentralized and emergent effects of such variation are, unfortunately, also difficult to systematically measure, analyze, and predict; modeling assumptions which are stringent enough to make analysis tractable frequently cannot be guaranteed at meaningful application scales, and longitudinal methods at such scales can require the capture and manipulation of impractically large amounts of data. This paper describes a new, scalable, and statistically robust approach for effective modeling, measurement, and analysis of large-scale performance variation in HPC systems. Our approach avoids the need to reason about complex distributions of runtimes among large numbers of individual application processes by focusing instead on the maximum length of distributed workload intervals. We describe this approach and its implementation in MPI which makes it applicable to a diverse set of HPC workloads. We also present evaluations of these techniques for quantifying and predicting performance variation carried out on large-scale computing systems, and discuss the strengths and limitations of the underlying modeling assumptions.

This report is an institutional record of experiments conducted to explore performance of a vendor installation of CephFS on the SNL stria cluster. Comparisons between CephFS, the Lustre parallel file system, and NFS were done using the IOR and MDTEST benchmarking tools, a test program which uses the SEACAS/Trilinos IOSS library, and the checkpointing activity performed by the LAMMPS molecular dynamics simulation.

Abstract not provided.

Abstract not provided.

Remote Direct Memory Access (RDMA) is an increasingly important technology in high-performance computing (HPC). RDMA provides low-latency, high-bandwidth data transfer between compute nodes. Additionally, it does not require explicit synchronization with the destination processor. Eliminating unnecessary synchronization can significantly improve the communication performance of large-scale scientific codes. A long-standing challenge presented by RDMA communication is mitigating the cost of registering memory with the network interface controller (NIC). Reusing memory once it is registered has been shown to significantly reduce the cost of RDMA communication. However, existing approaches for reusing memory rely on implicit memory semantics. In this paper, we introduce an approach that makes memory reuse semantics explicit by exposing a separate allocator for registered memory. The data and analysis in this paper yield the following contributions: (i) managing registered memory explicitly enables efficient reuse of registered memory; (ii) registering large memory regions to amortize the registration cost over multiple user requests can significantly reduce cost of acquiring new registered memory; and (iii) reducing the cost of acquiring registered memory can significantly improve the performance of RDMA communication. Reusing registered memory is key to high-performance RDMA communication. By making reuse semantics explicit, our approach has the potential to improve RDMA performance by making it significantly easier for programmers to efficiently reuse registered memory.

Coordinated checkpoint/restart is currently the dominant approach to mitigating the impact of failures on important scientific applications running on large-scale distributed systems. However, there is widespread evidence that coordinated checkpointing may no longer be viable on next-generation systems. Uncoordinated checkpoint/restart attempts to address the shortcomings of coordinated checkpoint/restart by allowing application processes to checkpoint their state independently. However, eliminating coordination may significantly degrade application performance. In this paper, we propose an approach that leverages existing coordination in important scientific applications to approximately coordinate checkpoints. Specifically, we propose to extend MPI implementations to force checkpoints to occur immediately after the completion of a collective operation. We evaluate the performance implications of this approach using an existing validated simulation framework. Our results demonstrate that approximately coordinated checkpointing can significantly improve application performance relative to totally uncoordinated checkpointing. We also show that forcing checkpoints to occur following a collective operation has a small impact on the nominal checkpoint interval for several important workloads. As a whole, the results presented in this paper demonstrate that approximately coordinated checkpointing may provide significant performance benefits without significantly increasing the cost of failure recovery.

Scientific data collections grow ever larger, both in terms of the size of individual data items and of the number and complexity of items. To use and manage them, it is important to directly address issues of robust and actionable provenance. We identify three key drivers as our focus: managing the size and complexity of metadata, lack of a priori information to match usage intents between publishers and consumers of data, and support for campaigns over collections of data driven by multi-disciplinary, collaborating teams. We introduce the Hoarde abstraction as an attempt to formalize a way of looking at collections of data to make them more tractable for later use. Hoarde leverages middleware and systems infrastructures for scientific and technical data management. Through the lens of a select group of challenging data usage scenarios, we discuss some of the aspects of implementation, usage, and forward portability of this new view on data management.

Composition of computational science applications into both ad hoc pipelines for analysis of collected or generated data and into well-defined and repeatable workflows is becoming increasingly popular. Meanwhile, dedicated high performance computing storage environments are rapidly becoming more diverse, with both significant amounts of non-volatile memory storage and mature parallel file systems available. At the same time, computational science codes are being coupled to data analysis tools which are not filesystem-oriented. In this paper, we describe how the FAODEL data management service can expose different available data storage options and mediate among them in both application- and FAODEL-directed ways. These capabilities allow applications to exploit their knowledge of the different types of data they may exchange during a workflow execution, and also provide FAODEL with mechanisms to proactively tune data storage behavior when appropriate. We describe the implementation of these capabilities in FAODEL and how they are used by applications, and present preliminary performance results demonstrating the potential benefits of our approach.

Composition of computational science applications, whether into ad hoc pipelines for analysis of simulation data or into well-defined and repeatable workflows, is becoming commonplace. In order to scale well as projected system and data sizes increase, developers will have to address a number of looming challenges. Increased contention for parallel filesystem bandwidth, accomodating in situ and ex situ processing, and the advent of decentralized programming models will all complicate application composition for next-generation systems. In this paper, we introduce a set of data services, Faodel, which provide scalable data management for workflows and composed applications. Faodel allows workflow components to directly and efficiently exchange data in semantically appropriate forms, rather than those dictated by the storage hierarchy or programming model in use. We describe the architecture of Faodel and present preliminary performance results demonstrating its potential for scalability in workflow scenarios.

Abstract not provided.

Recent high-performance computing (HPC) platforms such as the Trinity Advanced Technology System (ATS-1) feature burst buffer resources that can have a dramatic impact on an application’s I/O performance. While these non-volatile memory (NVM) resources provide a new tier in the storage hierarchy, developers must find the right way to incorporate the technology into their applications in order to reap the benefits. Similar to other laboratories, Sandia is actively investigating ways in which these resources can be incorporated into our existing libraries and workflows without burdening our application developers with excessive, platform-specific details. This FY18Q1 milestone summaries our progress in adapting the Sandia Parallel Aerodynamics and Reentry Code (SPARC) in Sandia’s ATDM program to leverage Trinity’s burst buffers for checkpoint/restart operations. We investigated four different approaches with varying tradeoffs in this work: (1) simply updating job script to use stage-in/stage out burst buffer directives, (2) modifying SPARC to use LANL’s hierarchical I/O (HIO) library to store/retrieve checkpoints, (3) updating Sandia’s IOSS library to incorporate the burst buffer in all meshing I/O operations, and (4) modifying SPARC to use our Kelpie distributed memory library to store/retrieve checkpoints. Team members were successful in generating initial implementation for all four approaches, but were unable to obtain performance numbers in time for this report (reasons: initial problem sizes were not large enough to stress I/O, and SPARC refactor will require changes to our code). When we presented our work to the SPARC team, they expressed the most interest in the second and third approaches. The HIO work was favored because it is lightweight, unobtrusive, and should be portable to ATS-2. The IOSS work is seen as a long-term solution, and is favored because all I/O work (including checkpoints) can be deferred to a single library.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Fault-tolerance poses a major challenge for future large-scale systems. Active research into coordinated, uncoordinated, and hybrid checkpointing systems has explored how the introduction of asynchrony can address anticipated scalability issues. While fully uncoordinated approaches have been shown to have significant delays, the degree of sychronization required to keep overheads low has not yet been significantly addressed. In this paper, we use a simulation-based approach to show the impact of synchronization on local checkpoint activity. Specifically, we show the degree of synchronization needed to keep the impacts of local checkpointing low is attainable with current technology for a number of key production HPC workloads. Our work provides a critical analysis and comparison of synchronization and local checkpointing. This enables users and system administrators to fine-tune the checkpointing scheme to the application and system characteristics available.

Abstract not provided.

Scientific workloads running on current extreme-scale systems routinely generate tremendous volumes of data for postprocessing. This data movement has become a serious issue due to its energy cost and the fact that I/O bandwidths have not kept pace with data generation rates. In situ analytics is an increasingly popular alternative in which post-simulation processing is embedded into an application, running as part of the same MPI job. This can reduce data movement costs but introduces a new potential source of interference for the application. Using a validated simulation-based approach, we investigate how best to mitigate the interference from time-shared in situ tasks for a number of key extreme-scale workloads. This paper makes a number of contributions. First, we show that the independent scheduling of in situ analytics tasks can significantly degradation application performance, with slowdowns exceeding 1000%. Second, we demonstrate that the degree of synchronization found in many modern collective algorithms is sufficient to significantly reduce the overheads of this interference to less than 10% in most cases. Finally, we show that many applications already frequently invoke collective operations that use these synchronizing MPI algorithms. Therefore, the syncronization introduced by these MPI collective algorithms can be leveraged to efficiently schedule analytics tasks with minimal changes to existing applications. This paper provides critical analysis and guidance for MPI users and developers on the importance of scheduling in situ analytics tasks. It shows the degree of synchronization needed to mitigate the performance impacts of these time-shared coupled codes and demonstrates how that synchronization can be realized in an extreme-scale environment using modern collective algorithms.

Next-generation applications increasingly rely on in situ analytics to guide computation, reduce the amount of I/O performed, and perform other important tasks. Scheduling where and when to run analytics is challenging, however. This paper quantifies the costs and benefits of different approaches to scheduling applications and analytics on nodes in large-scale applications, including space sharing, uncoordinated time sharing, and gang scheduled time sharing.

Next-generation systems face a wide range of new potential sources of application interference, including resilience actions, system software adaptation, and in situ analytics programs. In this paper, we present a new model for analyzing the performance of bulk-synchronous HPC applications based on the use of extreme value theory. After validating this model against both synthetic and real applications, the paper then uses both simulation and modeling techniques to profile next-generation interference sources and characterize their behavior and performance impact on a selection of HPC benchmarks, mini-applications, and applications. Lastly, this work shows how the model can be used to understand how current interference mitigation techniques in multi-processors work.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Relaxed synchronization offers the potential for maintaining application scalability, by allowing many processes to make independent progress when some processes suffer delays. Yet the benefits of this approach for important parallel workloads have not been investigated in detail. In this paper, we use a validated simulation approach to explore the noise-mitigation effects of idealized nonblocking collectives, in workloads where these collectives are a major contributor to total execution time. Although nonblocking collectives are unlikely to provide significant noise mitigation to applications in the low operating system noise environments expected in next-generation high-performance computing systems, we show that they can potentially improve application runtime with respect to other noise types.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Memory failures in future extreme scale applications are a significant concern in the high-performance computing community and have attracted much research attention. We contend in this paper that using application checkpoint data to detect memory failures has potential benefits and is preferable to examining application memory. To support this contention, we describe the application of machine learning techniques to evaluate the veracity of checkpoint data. Our preliminary results indicate that supervised decision tree machine learning approaches can effectively detect corruption in restart files, suggesting that future extreme-scale applications and systems may benefit from incorporating such approaches in order to cope with memory failues.

Abstract not provided.

Abstract not provided.

Exascale data environments are fast approaching, driven by diverse structured and unstructured data such as system and application telemetry streams, open-source information capture, and on-demand simulation output. Storage costs having plummeted, the question is now one of converting vast stores of data to actionable information. Complicating this problem are the low degrees of awareness across domain boundaries about what potentially useful data may exist, and write-once- read-never issues (data generation/collection rates outpacing data analysis and integration rates). Increasingly, technologists and researchers need to correlate previously unrelated data sources and artifacts to produce fused data views for domain-specific purposes. New tools and approaches for creating such views from vast amounts of data are vitally important to maintaining research and operational momentum. We propose to research and develop tools and services to assist in the creation, refinement, discovery and reuse of fused data views over large, diverse collections of heterogeneously structured data. We innovate in the following ways. First, we enable and encourage end-users to introduce customized index methods selected for local benefit rather than for global interaction (flexible multi-indexing). We envision rich combinations of such views on application data: views that span backing stores with different semantics, that introduce analytic methods of indexing, and that define multiple views on individual data items. We specifically decline to build a big fused database of everything providing a centralized index over all data, or to export a rigid schema to all comers as in federated query approaches. Second, we proactively advertise these application-specific views so that they may be programmatically reused and extended (data proactivity). Through this mechanism, both changes in state (new data in existing view collected) and changes in structure (new or derived view exists) are made known. Lastly, we embrace found data heterogeneity by coupling multi-indexing to backing stores with appropriate semantics (as opposed to a single store or schema).

Abstract not provided.

Abstract not provided.

Fault-tolerance has been identified as a major challenge for future extreme-scale systems. Current predictions suggest that, as systems grow in size, failures will occur more frequently. Because increases in failure frequency reduce the performance and scalability of these systems, significant effort has been devoted to developing and refining resilience mechanisms to mitigate the impact of failures. However, effective evaluation of these mechanisms has been challenging. Current systems are smaller and have significantly different architectural features (e.g., interconnect, persistent storage) than we expect to see in next-generation systems. To overcome these challenges, we propose the use of simulation. Simulation has been shown to be an effective tool for investigating performance characteristics of applications on future systems. In this work, we: identify the set of system characteristics that are necessary for accurate performance prediction of resilience mechanisms for HPC systems and applications; demonstrate how these system characteristics can be incorporated into an existing large-scale simulator; and evaluate the predictive performance of our modified simulator. We also describe how we were able to optimize the simulator for large temporal and spatial scales-allowing the simulator to run 4x faster and use over 100x less memory.

Much recent research has explored fault-tolerance mechanisms intended for current and future extreme-scale systems. Evaluations of the suitability of checkpoint-based solutions have typically been carried out using relatively uncomplicated computational kernels designed to measure floating point performance. More recent investigations have added scaled-down "proxy" applications to more closely match the composition and behavior of deployed ones. However, the information obtained from these studies (whether floating point performance or application runtime) is not necessarily of the most value in evaluating resilience strategies. We observe that even when using a more sophisticated metric, the information available from evaluating uncoordinated checkpointing using both microbenchmarks and proxy applications does not agree. This implies that not only might researchers be asking the wrong questions, but that the answers to the right ones might be unexpected and potentially misleading. We seek to open a discussion on whether benchmarks designed to provide predictable performance evaluations of HPC hardware and toolchains are providing the right feedback for the evaluation of fault-tolerance in these applications, and more generally on how benchmarking of resilience mechanisms ought to be approached in the exascale design space. © 2014 Springer-Verlag Berlin Heidelberg.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.