Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Parallel file systems”

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 91 records · Page 5

Toward a persistent event-streaming system for high-performance computing applications

High-performance computing (HPC) applications have traditionally relied on parallel file systems and file transfer services to manage data movement and storage. Alternative approaches have been proposed that use direct communications between application components, trading persistence and fault tolerance for speed. Event-driven architectures, as popularized in enterprise contexts, present a compelling middle ground, avoiding the performance cost and API constraints of parallel file systems while retaining persistence and offering impedance matching between application components. However, adapting streaming frameworks to HPC workloads requires addressing challenges unique to HPC systems. This paper investigates the potential for a streaming framework designed for HPC infrastructures and use cases. We introduce Mofka, a persistent event-streaming framework designed specifically for HPC environments. Mofka combines the capabilities of a traditional streaming service with optimizations tailored to the HPC context, such as support for massively multicore nodes, efficient scaling for large producer-consumer workflows, RDMA-enabled high-performance network communications, specialized network fabrics with multiple links per node, and efficient handling of large scientific data payloads. Built using the Mochi suite of HPC data service components, Mofka provides a lightweight, modular, and high-performance solution for persistent streaming in HPC systems. We present the architecture of Mofka and evaluate its performance against Kafka and Redpanda using benchmarks on diverse platforms, including Argonne's Polaris and Oak Ridge's Frontier supercomputers, showing up to 8× improvement in throughput in some scenarios. We then demonstrate its utility in several real-world applications: a tomographic reconstruction pipeline, a workflow for the discovery of metal-organic frameworks for carbon capture, and the instrumentation of Dask workflows for provenance tracking and performance analysis.

HPC↗

Comparative I/O workload characterization of two leadership class storage clusters. In: PDSW '15 Proceedings of the 10th Parallel Data Storage Workshop

The Oak Ridge Leadership Computing Facility (OLCF) is a leader in large-scale parallel file system development, design, deployment and continuous operation. For the last decade, the OLCF has designed and deployed two large center-wide parallel file systems. The first instantiation, Spider 1, served the Jaguar supercomputer and its predecessor, Spider 2, now serves the Titan supercomputer, among many other OLCF computational resources. The OLCF has been rigorously collecting file and storage system statistics from these Spider systems since their transition to production state. In this paper we present the collected I/O workload statistics from the Spider 2 system and compare it to the Spider 1 data. Our analysis show that the Spider 2 workload is more more write-heavy I/O compared to Spider 1 (75% vs. 60%, respectively). The data also show the OLCF storage policies such as periodic purges are effectively managing the capacity resource of Spider 2. Furthermore, due to improvements in tdm_multipath and ib_srp software, we are utilizing the Spider 2 system bandwidth and latency resources more effectively. The Spider 2 bandwidth usage statistics shows that our system is working within the design specifications. However, it is also evident that our scientific applications can be more effectively served by a burst buffer storage layer. All the data has been collected by monitoring tools developed for the Spider ecosystem. We believe the observed data set and insights will help us better design the next-generation Spider file and storage system. It will also be helpful to the larger community for building more effective large-scale file and storage systems.

Gunasekaran, Raghul↗

Layout-aware I/O Scheduling for terabits data movement

Many science facilities, such as the Department of Energy's Leadership Computing Facilities and experimental facilities including the Spallation Neutron Source, Stanford Linear Accelerator Center, and Advanced Photon Source, produce massive amounts of experimental and simulation data. These data are often shared among the facilities and with collaborating institutions. Moving large datasets over the wide-area network (WAN) is a major problem inhibiting collaboration. Next-generation, terabit-networks will help alleviate the problem, however, the parallel storage systems on the endsystem hosts at these institutions can become a bottleneck for terabit data movement. The parallel storage system (PFS) is shared by simulation systems, experimental systems, analysis and visualization clusters, in addition to wide-area data movers. These competing uses often induce temporary, but significant, I/O load imbalances on the storage system, which impact the performance of all the users. The problem is a serious concern because some resources are more expensive (e.g. super computers) or have time-critical deadlines (e.g. experimental data from a light source), but parallel file systems handle all requests fairly even if some storage servers are under heavy load. This paper investigates the problem of competing workloads accessing the parallel file system and how the performance of wide-area data movement can be improved in these environments. First, we study the I/O load imbalance problems using actual I/O performance data collected from the Spider storage system at the Oak Ridge Leadership Computing Facility. Second, we present I/O optimization solutions with layout-awareness on end-system hosts for bulk data movement. With our evaluation, we show that our I/O optimization techniques can avoid the I/O congested disk groups, improving storage I/O times on parallel storage systems for terabit data movement.

Kim, Youngjae↗

Layout-Aware I/O Scheduling for Terabits Data Movement

Many science facilities, such as the Department of Energy s Leadership Computing Facilities and experimental facilities including the Spallation Neutron Source, Stanford Linear Accelerator Center, and Advanced Photon Source, produce massive amounts of experimental and simulation data. These data are often shared among the facilities and with collaborating institutions. Moving large datasets over the wide- area network (WAN) is a major problem inhibiting collaboration. Next- generation, terabit-networks will help alleviate the problem, however, the parallel storage systems on the end-system hosts at these institutions can become a bottleneck for terabit data movement. The parallel storage system (PFS) is shared by simulation systems, experimental systems, analysis and visualization clusters, in addition to wide-area data movers. These competing uses often induce temporary, but significant, I/O load imbalances on the storage system, which impact the performance of all the users. The problem is a serious concern because some resources are more expensive (e.g. super computers) or have time-critical deadlines (e.g. experimental data from a light source), but parallel file systems handle all requests fairly even if some storage servers are under heavy load. This paper investigates the problem of competing workloads accessing the parallel file system and how the performance of wide-area data movement can be improved in these environments. First, we study the I/O load imbalance problems using actual I/O performance data collected from the Spider storage system at the Oak Ridge Leadership Computing Facility. Second, we present I/O optimization solutions with layout-awareness on end-system hosts for bulk data movement. With our evaluation, we show that our I/O optimization techniques can avoid the I/O congested disk groups, improving storage I/O times on parallel storage systems for terabit data movement.

I/O and Storage↗

FitCache: A Transparent Drop-In Framework for Multi-Tier Caching to Accelerate Distributed Deep Learning Workloads

Training in Deep learning (DL) remains highly compute- and data-intensive, with I/O becoming a critical bottleneck as models and datasets scale. Recent studies report that data loading can dominate training time, especially on large-scale HPC systems with shared parallel file systems (PFS). Existing caching approaches either rely on single-tier designs or require intrusive modifications to training pipelines, limiting their portability and effectiveness. In this work, we present FitCache, a transparent drop-in framework for multi-tier caching to accelerate distributed DL training by coordinating fast local memory (e.g., DRAM, Persistent Memory (PMem)) and NVMe as hierarchical caches atop PFS. Our design adapts to hardware diversity, i.e., if NVMe is missing, memory transparently acts as a caching tier, ensuring stable performance. FitCache transparently intercepts I/O requests and issues concurrent fetches across all tiers, returning data from the fastest responder without centralized metadata or static redirection paths. FitCache adapts to dynamic workloads and heterogeneous clusters while maintaining POSIX compatibility. Experiments on Frontier (2048 GPUs) and smaller research clusters show that FitCache reduces training time by up to 40% and per-batch I/O latency by up to 71.6% compared to Lustre Orion PFS, offering a drop-in solution for scalable DL training.

Hu, Guangxing [ORNL] (ORCID:0009000283203614)↗

Understanding and Improving High-Performance I/O Subsystems

This research program has been conducted in the framework of the NASA Earth and Space Science (ESS) evaluations led by Dr. Thomas Sterling. In addition to the many important research findings for NASA and the prestigious publications, the program has helped orienting the doctoral research program of two students towards parallel input/output in high-performance computing. Further, the experimental results in the case of the MasPar were very useful and helpful to MasPar with which the P.I. has had many interactions with the technical management. The contributions of this program are drawn from three experimental studies conducted on different high-performance computing testbeds/platforms, and therefore presented in 3 different segments as follows: 1. Evaluating the parallel input/output subsystem of a NASA high-performance computing testbeds, namely the MasPar MP- 1 and MP-2; 2. Characterizing the physical input/output request patterns for NASA ESS applications, which used the Beowulf platform; and 3. Dynamic scheduling techniques for hiding I/O latency in parallel applications such as sparse matrix computations. This study also has been conducted on the Intel Paragon and has also provided an experimental evaluation for the Parallel File System (PFS) and parallel input/output on the Paragon. This report is organized as follows. The summary of findings discusses the results of each of the aforementioned 3 studies. Three appendices, each containing a key scholarly research paper that details the work in one of the studies are included.

El-Ghazawi, Tarek A.↗

Scalable I/O Systems via Node-Local Storage: Approaching 1 TB/sec File I/O

In the race to PetaFLOP-speed supercomputing systems, the increase in computational capability has been accompanied by corresponding increases in CPU count, total RAM, and storage capacity. However, a proportional increase in storage bandwidth has lagged behind. In order to improve system reliability and to reduce maintenance effort for modern large-scale systems, system designers have opted to remove node-local storage from the compute nodes. Today's multi-TeraFLOP supercomputers are typically attached to parallel file systems that provide only tens of GBs/s of I/O bandwidth. As a result, such machines have access to much less than 1GB/s of I/O bandwidth per TeraFLOP of compute power, which is below the generally accepted limit required for a well-balanced system. In a many ways, the current I/O bottleneck limits the capabilities of modern supercomputers, specifically in terms of limiting their working sets and restricting fault tolerance techniques, which become critical on systems consisting of tens of thousands of components. This paper resolves the dilemma between high performance and high reliability by presenting an alternative system design which makes use of node-local storage to improve aggregate system I/O bandwidth. In this work, we focus on the checkpointing use-case and present an experimental evaluation of the Scalable Checkpoint/Restart (SCR) library, a new adaptive checkpointing library that uses node-local storage to significantly improve the checkpointing performance of large-scale supercomputers. Experiments show that SCR achieves unprecedented write speeds, reaching a measured 700GB/s of aggregate bandwidth on 8,752 processors and an estimated 1TB/s for a similarly structured machine of 12,500 processors. This corresponds to a speedup of over 70x compared to the bandwidth provided by the 10GB/s parallel file system the cluster uses. Further, SCR can adapt to an environment in which there is wide variation in performance or capacity among the individual node-local storage elements.

99 GENERAL AND MISCELLANEOUS↗

The Scalable Checkpoint/Restart Library

The Scalable Checkpoint/Restart (SCR) library provides an interface that codes may use to worite our and read in application-level checkpoints in a scalable fashion. In the current implementation, checkpoint files are cached in local storage (hard disk or RAM disk) on the compute nodes. This technique provides scalable aggregate bandwidth and uses storage resources that are fully dedicated to the job. This approach addresses the two common drawbacks of checkpointing a large-scale application to a shared parallel file system, namely, limited bandwidth and file system contention. In fact, on current platforms, SCR scales linearly with the number of compute nodes. It has been benchmarked as high as 720GB/s on 1094 nodes of Atlas, which is nearly two orders of magnitude faster thanthe parallel file system.

ECP↗

HPC Global File System Performance Analysis Using A Scientific-Application Derived Benchmark

With the exponential growth of high-fidelity sensor and simulated data, the scientific community is increasingly reliant on ultrascale HPC resources to handle its data analysis requirements. However, to use such extreme computing power effectively, the I/O components must be designed in a balanced fashion, as any architectural bottleneck will quickly render the platform intolerably inefficient. To understand I/O performance of data-intensive applications in realistic computational settings, we develop a lightweight, portable benchmark called MADbench2, which is derived directly from a large-scale Cosmic Microwave Background (CMB) data analysis package. Our study represents one of the most comprehensive I/O analyses of modern parallel file systems, examining a broad range of system architectures and configurations, including Lustre on the Cray XT3, XT4, and Intel Itanium2 clusters; GPFS on IBM Power5 and AMD Opteron platforms; a BlueGene/P installation using GPFS and PVFS2 file systems; and CXFS on the SGI Altix\-3700. We present extensive synchronous I/O performance data comparing a number of key parameters including concurrency, POSIX- versus MPI-IO, and unique-versus shared-file accesses, using both the default environment as well as highly-tuned I/O parameters. Finally, we explore the potential of asynchronous I/O and show that only the two of the nine evaluated systems benefited from MPI-2's asynchronous MPI-IO. On those systems, experimental results indicate that the computational intensity required to hide I/O effectively is already close to the practical limit of BLAS3 calculations. Overall, our study quantifies vast differences in performance and functionality of parallel file systems across state-of-the-art platforms -- showing I/O rates that vary up to 75x on the examined architectures -- while providing system designers and computational scientists a lightweight tool for conducting further analysis.

97 MATHEMATICS AND COMPUTING↗

SDS: A Framework for Scientific Data Services

Large-scale scientific applications typically write their data to parallel file systems with organizations designed to achieve fast write speeds. Analysis tasks frequently read the data in a pattern that is different from the write pattern, and therefore experience poor I/O performance. In this paper, we introduce a prototype framework for bridging the performance gap between write and read stages of data access from parallel file systems. We call this framework Scientific Data Services, or SDS for short. This initial implementation of SDS focuses on reorganizing previously written files into data layouts that benefit read patterns, and transparently directs read calls to the reorganized data. SDS follows a client-server architecture. The SDS Server manages partial or full replicas of reorganized datasets and serves SDS Clients' requests for data. The current version of the SDS client library supports HDF5 programming interface for reading data. The client library intercepts HDF5 calls and transparently redirects them to the reorganized data. The SDS client library also provides a querying interface for reading part of the data based on user-specified selective criteria. We describe the design and implementation of the SDS client-server architecture, and evaluate the response time of the SDS Server and the performance benefits of SDS.

59 BASIC BIOLOGICAL SCIENCES↗

Adaptive multi-level checkpointing

In some examples, with respect to adaptive multi-level checkpointing, a transfer parameter associated with transfer of checkpoint data from a node-local storage to a parallel file system may be ascertained for the checkpoint data stored in the node-local storage. The transfer parameter may be compared to a specified transfer parameter threshold. A determination may be made, based on the comparison of the transfer parameter to the specified transfer parameter threshold, as to whether to transfer the checkpoint data from the node-local storage to the parallel file system.

97 MATHEMATICS AND COMPUTING↗

IOPro: a parallel I/O profiling and visualization framework for high-performance storage systems

Efficient execution of large-scale scientific applications requires high-performance computing systems designed to meet the I/O requirements. To achieve high-performance, such data-intensive parallel applications use a multi-layer layer I/O software stack, which consists of high-level I/O libraries such as PnetCDF and HDF5, the MPI library, and parallel file systems. To design efficient parallel scientific applications, understanding the complicated flow of I/O operations and the involved interactions among the libraries is quintessential. Such comprehension helps identify I/O bottlenecks and thus exploits the potential performance in different layers of the storage hierarchy. To profile the performance of individual components in the I/O stack and to understand complex interactions among them, we have implemented a GUI-based integrated profiling and analysis framework, IOPro. IOPro automatically generates an instrumented I/O stack, runs applications on it, and visualizes detailed statistics based on the user-specified metrics of interest. We present experimental results from two different real-life applications and show how our framework can be used in practice. By generating an end-to-end trace of the whole I/O stack and pinpointing I/O interference, IOPro aids in understanding I/O behavior and improving the I/O performance significantly.

97 MATHEMATICS AND COMPUTING↗

I/O Bottleneck Detection and Tuning: Connecting the Dots using Interactive Log Analysis

Using parallel file systems efficiently is a tricky problem due to inter-dependencies among multiple layers of I/O software, including high-level I/O libraries (HDF5, netCDF, etc.), MPI-IO, POSIX, and file systems (GPFS, Lustre, etc.). Profiling tools such as Darshan collect traces to help understand the I/O performance behavior. However, there are significant gaps in analyzing the collected traces and then applying tuning options offered by various layers of I/O software. Seeking to connect the dots between I/O bottleneck detection and tuning, we propose DXT Explorer, an interactive log analysis tool. In this paper, we present a case study using our interactive log analysis tool to identify and apply various I/O optimizations. We report an evaluation of performance improvement achieved for four I/O kernels extracted from science applications.

Bez, Jean Luca↗

Large Scale Caching and Streaming of Training Data for Online Deep Learning

The training of deep neural network models on large data remains a difficult problem, despite progress towards scalable techniques. In particular, there is a mismatch between the random but predetermined order in which AI flows select training samples and the streaming I/O patterns for which traditional HPC data storage (e.g., parallel file systems) are designed. In addition, as more data are obtained, it is feasible neither simply to train learning models incrementally, due to catastrophic forgetting (i.e., bias towards new samples), nor to train frequently from scratch, due to prohibitive time and/or resource constraints. In this paper, we study data management techniques that combine caching and streaming with rehearsal support in order to enable efficient access to training samples in both offline training and continual learning. We revisit state-of-art streaming approaches based on data pipelines that transparently handle prefetching, caching, shuffling, and data augmentation, and discuss the challenges and opportunities that arise when combining these methods with data-parallel training techniques. We also report on preliminary experiments that evaluate the I/O overheads involved in accessing the training samples from a parallel file system (PFS) under several concurrency scenarios, highlighting the impact of the PFS on the design of the data pipelines.

data pipelines↗

Containers for Massive Ensemble of I/O Bound Hierarchical Coupled Simulations

We present our experience using containers to scale up a massive ensemble of coupled I/O bound workloads on the NERSC Cori supercomputer. We describe the design of a hierarchical simulation structure using the Integrated Plasma Simulator (IPS) that enables the flexible execution of coupled simulations at the system, node, and core level using the same coupling abstraction and API. The hierarchical design allows for the node-level execution to be efficiently executed using containers while not impacting the structure of the simulation at the system level. We demonstrate the viability of the approach by presenting experimental results from applications in coupled fusion plasma simulations that illustrate the performance impact of using containers to deploy the node-level workloads, in conjunction with the user mountable XFS file systems to ameliorate the load on the Lustre parallel file system. We also present results from production runs showing the ability of the ensemble simulations to scale to hundreds of Cori Haswell nodes, with little or no overhead.

Elwasif, Wael↗

Comparison of leading parallel NAS file systems on commodity hardware

High performance computing has experienced tremendous gains in system performance over the past 20 years. Unfortunately other system capabilities, such as file I/O, have not grown commensurately. In this activity, we present the results of our tests of two leading file systems (GPFS and Lustre) on the same physical hardware. This hardware is the standard commodity storage solution in use at LLNL and, while much smaller in size, is intended to enable us to learn about differences between the two systems in terms of performance, ease of use and resilience. This work represents the first hardware consistent study of the two leading file systems that the authors are aware of.

97 MATHEMATICS AND COMPUTING↗

Optimizing End-to-End Big Data Transfers over Terabits Network Infrastructure

While future terabit networks hold the promise of significantly improving big-data motion among geographically distributed data centers, significant challenges must be overcome even on today's 100 gigabit networks to realize end-to-end performance. Multiple bottlenecks exist along the end-to-end path from source to sink, for instance, the data storage infrastructure at both the source and sink and its interplay with the wide-area network are increasingly the bottleneck to achieving high performance. In this study, we identify the issues that lead to congestion on the path of an end-to-end data transfer in the terabit network environment, and we present a new bulk data movement framework for terabit networks, called LADS. LADS exploits the underlying storage layout at each endpoint to maximize throughput without negatively impacting the performance of shared storage resources for other users. LADS also uses the Common Communication Interface (CCI) in lieu of the sockets interface to benefit from hardware-level zero-copy, and operating system bypass capabilities when available. It can further improve data transfer performance under congestion on the end systems using buffering at the source using flash storage. With our evaluations, we show that LADS can avoid congested storage elements within the shared storage resource, improving input/output bandwidth, and data transfer rates across the high speed networks. We also investigate the performance degradation problems of LADS due to I/O contention on the parallel file system (PFS), when multiple LADS tools share the PFS. We design and evaluate a meta-scheduler to coordinate multiple I/O streams while sharing the PFS, to minimize the I/O contention on the PFS. Finally, with our evaluations, we observe that LADS with meta-scheduling can further improve the performance by up to 14 percent relative to LADS without meta-scheduling.

97 MATHEMATICS AND COMPUTING↗