Search NASA⌕ Search

SEARCH · Search NASA

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
NASA NTRS1996

Database Reorganization in Parallel Disk Arrays with I/O Service Stealing

We present a model for data reorganization in parallel disk systems that is geared towards load balancing in an environment with periodic access patterns. Data reorganization is performed by disk cooling, i.e. migrating files or extents from the hottest disks to the coldest ones. We develop an approximate queueing model for determining the effective arrival rates of cooling requests and discuss its use in assessing the costs versus benefits of cooling.

Zabback, Peter↗
DOE OSTI2021

Scalable filesystem enumeration and metadata operations

Systems, apparatus, and methods are disclosed for performing scalable operations in a file system, including POSIX-like file systems. Metadata entries in a namespace or directory tree are sharded across multiple file metadata servers. An enumeration operation, such as listing a directory, is parallelized across the multiple file metadata servers, while retaining standard functionality transparently to clients. Other enumeration operations include no-output operations such as changing file attributes or deleting a file, and cumulative operations such as counting total disk space usage. The parallelization is compatible with tree-level parallelization and storage-level parallelization. Disclosed technologies can be applied to other fields requiring scalable enumeration, such as database and network applications.

Grider, Gary A.↗
DOE OSTI2020

How fast can one resize a distributed file system?

Efficient resource utilization becomes a major concern as large-scale distributed computing infrastructures keep growing in size. Malleability, the possibility for resource managers to dynamically increase or decrease the amount of resources allocated to a job, is a promising way to save energy and costs. However, state-of-the-art parallel and distributed storage systems have not been designed with malleability in mind. The reason is mainly the supposedly high cost of data transfers required by resizing operations. Nevertheless, as network and storage technologies evolve, old assumptions about potential bottlenecks can be revisited. In this study, we evaluate the viability of malleability as a design principle for a distributed storage system. We specifically model the minimal duration of the commission and decommission operations. To show how our models can be used in practice, we evaluate the performance of these operations in HDFS, a relevant state-of-the-art distributed file system. We show that the existing decommission mechanism of HDFS is good when the network is the bottleneck, but can be accelerated by up to a factor 3 when storage is the limiting factor. We also show that the commission in HDFS can be substantially accelerated. With the highlights provided by our model, we suggest improvements to speed both operations in HDFS. We discuss how the proposed models can be generalized for distributed file systems with different assumptions and what perspectives are open for the design of efficient malleable distributed file systems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Scalable augmented enumeration and metadata operations for large filesystems

Systems, apparatus, and methods are disclosed for performing scalable operations in a file system. Metadata entries in a namespace or directory tree are sharded across multiple file metadata servers. An augmented enumeration operation, such as listing a directory, is parallelized across the multiple file metadata servers, transparently to clients. Exemplary augmentation features can include filtering and sorting. Augmentation features can be executed concurrently with enumeration, prior to enumeration, after enumeration, or as a combination of these, and can utilize pre-built index structures or holding structures for intermediate results. Augmented enumeration operations can also include no-output operations such as changing file attributes or deleting a file, and cumulative operations such as counting total disk space usage. The parallelization is compatible with tree-level parallelization and storage-level parallelization. Disclosed technologies can be applied to other fields requiring scalable enumeration, such as database and network applications.

Grider, Gary A.↗
DOE OSTI2021

Improving I/O Performance for Exascale Applications through Online Data Layout Reorganization

The applications being developed within the U.S. Exascale Computing Project (ECP) to run on imminent Exascale computers will generate scientific results with unprecedented fidelity and record turn-around time. Many of these codes are based on particle-mesh methods and use advanced algorithms, especially dynamic load-balancing and mesh-refinement, to achieve high performance on Exascale machines. Yet, as such algorithms improve parallel application efficiency, they raise new challenges for I/O logic due to their irregular and dynamic data distributions. Thus, while the enormous data rates of Exascale simulations already challenge existing file system write strategies, the need for efficient read and processing of generated data introduces additional constraints on the data layout strategies that can be used when writing data to secondary storage. We review these I/O challenges and introduce two online data layout reorganization approaches for achieving good tradeoffs between read and write performance. We demonstrate the benefits of using these two approaches for the ECP particle-in-cell simulation WarpX, which serves as a motif for a large class of important Exascale applications. Here, we show that by understanding application I/O patterns and carefully designing data layouts we can increase read performance by more than 80 percent.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Evolution of the ROOT Tree I/O

The ROOT TTree data format encodes hundreds of petabytes of High Energy and Nuclear Physics events. Its columnar layout drives rapid analyses, as only those parts (“branches”) that are really used in a given analysis need to be read from storage. Its unique feature is the seamless C++ integration, which allows users to directly store their event classes without explicitly defining data schemas. In this contribution, we present the status and plans of the future ROOT 7 event I/O. Along with the ROOT 7 interface modernization, we aim for robust, where possible compile-time safe C++ interfaces to read and write event data. On the performance side, we show first benchmarks using ROOT’s new experimental I/O subsystem that combines the best of TTrees with recent advances in columnar data formats. A core ingredient is a strong separation of the high-level logical data layout (C++ classes) from the low-level physical data layout (storage backed nested vectors of simple types). We show how the new, optimized physical data layout speeds up serialization and deserialization and facilitates parallel, vectorized and bulk operations. This lets ROOT I/O run optimally on the upcoming ultra-fast NVRAM storage devices, as well as file-less storage systems such as object stores.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2019

Parallel Event Selection on HPC Systems

In their recent measurement of the neutrino oscillation parameters,NOvA uses a sample of approximately 25 million reconstructed spills to searchfor electron-neutrino appearance events. These events are stored in an n-tupleformat, in 250 thousand ROOT files. File sizes range from a few hundred KiB toa few MiB; the full dataset is approximately 1.4 TiB. These millions of eventsare reduced to a few tens of events by the application of strict event selectioncriteria, and then summarized by a handful of numbers each, which are used inthe extraction of the neutrino oscillation parameters.The NOvA event selection code is currently a serial C++ program that readsthese n-tuples. The current table data format and organization and the selection/reduction processing involved provides us with an opportunity to explorealternate approaches to represent the data and implement the processing. Werepresent our n-tuple data in HDF5 format that is optimized for the HPC environmentand which allows us to use the machine’s high-performance parallelI/O capabilities. We use MPI, numpy and h5py to implement our approach andcompare the performance with the existing approach. We study the performanceimplications of using thousands of small files of different sizes as comparedwith one large file using HPC resources. This work has been done as part of theSciDAC project, “HEP analytics on HPC” in collaboration with the ASCR teamsat ANL and LBNL.

Paterno, Marc↗
DOE OSTI2023

ORNL_AISD_NiPt_108atoms

This dataset describes the nickel-platinum (NiPt) solid solution binary alloy, where the two constituent elements nickel (Ni) and platinum (Pt) are randomly placed on the face centered cubic (FCC) crystal structure, with the lattice constant of 3.840 angstroms. The dataset comprises data for crystal structures with 108 atoms with 1,900 configurations. The data set was generated for concentrations ranging from 0at% of Pt to 100at% of Pt in the NiPt binary system, with increasing the concentration of Pt in the system every 5at%. For each one of the chemical compositions, 100 random configurations were generated, each with a different random seed. Each of the output files contains the mass, type, atomic coordinates, energy per atom, and forces in x, y, and z directions respectively. For each atomic configuration, the output was collected every 150 steps during the minimization stage and every 1000 steps during the replica exchange stage. Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) [1], which is a molecular dynamics code, was used to generate data for NiPt alloy. The simulation used the interatomic potential for NiPt binary system 'MEAM_LAMMPS_KimSeolJi_2017_PtNi__MO_020840179467_001' [3] from the OpenKIM library (Open Knowledgebase of Interatomic Models) [2]. This potential was developed based on the second nearest-neighbor modified embedded-atom method (2NN MEAM). The simulation process begins with the generation of the random NiPt structure and follows with the short minimization and replica exchange simulation. The minimization procedure adjusts atomic coordinates and performs energy minimization, which typically leads to a local potential energy minimum. The method used for the minimization was the conjugate gradient algorithm. A short replica exchange (parallel tempering) simulation involves four replicas (ensembles) of a system and follows the minimization stage. Multiple snapshots of the configuration were collected during the minimization and replica exchange stages. NiPt alloy is interesting due to its magnetic and charge transfer properties [4]. The data is provided in a compressed zipped folders atoms108.zip. The zipped folder contains the data structured in the following way: - Ni_ground_state.cfg --> atomic configuration for the pure nickel - Pt_ground_state.cfg --> atomic configuration for the pure platinum - Pt#_filtered --> folders containing atomic configurations for #at% concentration of platinum. The folder contains 100 atomic configurations, each saved in a subfolder - Each subfolder named config* is associated with a specific atomic configuration. Each of these subfolders contains files with .cfg format, corresponding to outputs for each atomic configuration The total number of atomic configurations contained in atoms108.zip is 66,132. This dataset is an extension to the dataset ORNL_AISD_NiPt [5] that has been previously released with crystal structures of 256 atoms, 864 atoms, and 2,048 atoms, with the same methodology for data collection. References [1] https://www.lammps.org/ [2] https://openkim.org/ [3] https://openkim.org/id/MEAM_LAMMPS_KimSeolJi_2017_PtNi__MO_020840179467_001 [4] El-Gendy, Ahmed A. and Hampel, Silke and Büchner, Bernd and Klingeler, Rüdiger, Tuneable magnetic properties of carbon-shielded NiPt-nanoalloys, RSC Adv., volume 6, issue 57, pages 52427-52433, 2016, The Royal Society of Chemistry, doi:10.1039/C6RA05910D [5] M. Karabin, M. Lupo Pasini, and M. Eisenbach. ORNL_AISD_NiPt. United States: N. p., 2023. Web. doi:10.13139/OLCF/1958172.

36 MATERIALS SCIENCE↗
NASA NTRS1999

3D Multistage Simulation of Each Component of the GE90 Turbofan Engine

A 3D multistage simulation of each component of the GE90 Turbofan engine has been made. This includes 49 blade rows. A coupled simulation of all blade rows will be made very soon. The simulation is running using two levels of parallelism. The first level is on a blade row basis with information shared using files. The second level is using a grid domain decomposition with information shared using MPI. Timings will be shown for running on the SP2, an SGI Origin and a distributed system of HP workstations. On the HP workstations, the CHIMP version of MPI is used, with queuing supplied by LSF (Load Sharing Facility). A script-based control system is used to ensure reliability. An MPEG movie illustrating the flow simulation of the engine has been created using PV3, a parallel visualization library created by Bob Haimes of MIT. PVM is used to create a virtual machine from 10 HP workstations and display on an SGI workstation. A representative component simulation will be compared to rig data to demonstrate its usefulness in turbomachinery design and analysis.

Turner, Mark↗
DOE OSTI2023

ORNL_AISD_NiPt

This dataset describes the nickel-platinum (NiPt) solid solution binary alloy, where the two constituent elements nickel (Ni) and platinum (Pt) are randomly placed on the face centered cubic (FCC) crystal structure, with the lattice constant of 3.840 angstroms. The dataset comprises data for three different sizes of the crystal structure: 256 atoms, 864 atoms, and 2,048 atoms, each of which contains 1900 configurations. For each size of the crystal structure, the data set was generated for concentrations ranging from 0at% of Pt to 100at% of Pt in the NiPt binary system, with increasing the concentration of Pt in the system every 5at%. For each one of the chemical compositions, 100 random configurations were generated, each with a different random seed. Each of the output files contains the mass, type, atomic coordinates, energy per atom, and forces in x, y, and z directions respectively. For each atomic configuration, the output was collected every 150 steps during the minimization stage and every 1000 steps during the replica exchange stage. Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) [1], which is a molecular dynamics code, was used to generate data for NiPt alloy. The simulation used the interatomic potential for NiPt binary system MEAM_LAMMPS_KimSeolJi_2017_PtNi__MO_020840179467_001 [3] from the OpenKIM library (Open Knowledgebase of Interatomic Models) [2]. This potential was developed based on the second nearest-neighbor modified embedded-atom method (2NN MEAM). The simulation process begins with the generation of the random NiPt structure and follows with the short minimization and replica exchange simulation. The minimization procedure adjusts atomic coordinates and performs energy minimization, which typically leads to a local potential energy minimum. The method used for the minimization was the conjugate gradient algorithm. A short replica exchange (parallel tempering) simulation involves four replicas (ensembles) of a system and follows the minimization stage. Multiple snapshots of the configuration were collected during the minimization and replica exchange stages. NiPt alloy is interesting due to its magnetic and charge transfer properties [4]. The data is provided in three compressed zipped folders: atoms256.zip, atoms864.zip, atoms2048.zip Each zipped folder contains the data that describes crystals of size 256 atoms, 864 atoms, and 2,048 atoms respectively. Each one of the three zipped folders contains the data structured in the following way: -Ni_ground_state.cfg --> atomic configuration for the pure nickel -Pt_ground_state.cfg --> atomic configuration for the pure platinum -Pt#_filtered --> folders containing atomic configurations for #at% concentration of platinum. The folder contains 100 atomic configurations, each saved in a subfolder. Each subfolder named config* is associated with a specific atomic configuration. Each of these subfolders contains files with .cfg format, corresponding to outputs for each atomic configuration The total number of atomic configurations contained in atoms256.zip is 65,046. The total number of atomic configurations contained in atoms864.zip is 63,936. The total number of atomic configurations contained in atoms2048.zip is 61,997. The total number of atomic configurations spanned by the entire dataset is 190,979. References [1] https://www.lammps.org/ [2] https://openkim.org/ [3] https://openkim.org/id/MEAM_LAMMPS_KimSeolJi_2017_PtNi__MO_020840179467_001 [4] El-Gendy, Ahmed A. and Hampel, Silke and Büccchner, Bernd and Klingeler, Rüdiger, Tuneable magnetic properties of carbon-shielded NiPt-nanoalloys, RSC Adv., volume 6, issue 57, pages 52427-52433, 2016, The Royal Society of Chemistry, doi:10.1039/C6RA05910D

36 MATERIALS SCIENCE↗
DOE OSTI2018

Scalable Data Management of the Uintah Simulation Framework for Next-Generation Engineering Problems with Radiation

The need to scale next-generation industrial engineering problems to the largest computational platforms presents unique challenges. This paper focuses on data management related problems faced by the Uintah simulation framework at a production scale of 260K processes. Uintah provides a highly scalable asynchronous many-task runtime system, which in this work is used for the modeling of a 1000 megawatt electric (MWe) ultra-supercritical (USC) coal boiler. At 260K processes, we faced both parallel I/O and visualization related challenges, e.g., the default file-per-process I/O approach of Uintah did not scale on Mira. In this paper we present a simple to implement, restructuring based parallel I/O technique. We impose a restructuring step that alters the distribution of data among processes. The goal is to distribute the dataset such that each process holds a larger chunk of data, which is then written to a file independently. This approach finds a middle ground between two of the most common parallel I/O schemes–file per process I/O and shared file I/O–in terms of both the total number of generated files, and the extent of communication involved during the data aggregation phase. To address scalability issues when visualizing the simulation data, we developed a lightweight renderer using OSPRay, which allows scientists to visualize the data interactively at high quality and make production movies. Finally, this work presents a highly efficient and scalable radiation model based on the sweeping method, which significantly outperforms previous approaches in Uintah, like discrete ordinates. The integrated approach allowed the USC boiler problem to run on 260K CPU cores on Mira.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Profiling the BLAST bioinformatics application for load balancing on high-performance computing clusters

Abstract Background The Basic Local Alignment Search Tool (BLAST) is a suite of commonly used algorithms for identifying matches between biological sequences. The user supplies a database file and query file of sequences for BLAST to find identical sequences between the two. The typical millions of database and query sequences make BLAST computationally challenging but also well suited for parallelization on high-performance computing clusters. The efficacy of parallelization depends on the data partitioning, where the optimal data partitioning relies on an accurate performance model. In previous studies, a BLAST job was sped up by 27 times by partitioning the database and query among thousands of processor nodes. However, the optimality of the partitioning method was not studied. Unlike BLAST performance models proposed in the literature that usually have problem size and hardware configuration as the only variables, the execution time of a BLAST job is a function of database size, query size, and hardware capability. In this work, the nucleotide BLAST application BLASTN was profiled using three methods: shell-level profiling with the Unix “time” command, code-level profiling with the built-in “profiler” module, and system-level profiling with the Unix “gprof” program. The runtimes were measured for six node types, using six different database files and 15 query files, on a heterogeneous HPC cluster with 500+ nodes. The empirical measurement data were fitted with quadratic functions to develop performance models that were used to guide the data parallelization for BLASTN jobs. Results Profiling results showed that BLASTN contains more than 34,500 different functions, but a single function, RunMTBySplitDB, takes 99.12% of the total runtime. Among its 53 child functions, five core functions were identified to make up 92.12% of the overall BLASTN runtime. Based on the performance models, static load balancing algorithms can be applied to the BLASTN input data to minimize the runtime of the longest job on an HPC cluster. Four test cases being run on homogeneous and heterogeneous clusters were tested. Experiment results showed that the runtime can be reduced by 81% on a homogeneous cluster and by 20% on a heterogeneous cluster by re-distributing the workload. Discussion Optimal data partitioning can improve BLASTN’s overall runtime 5.4-fold in comparison with dividing the database and query into the same number of fragments. The proposed methodology can be used in the other applications in the BLAST+ suite or any other application as long as source code is available.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2021

Battle of the Defaults: Extracting Performance Characteristics of HDF5 under Production Load

Popular parallel I/O libraries, such as HDF5, provide tuning parameters to obtain superior performance. However, the selection of effective parameters on production systems is complex due to the interdependence of I/O software and file system layers. Hence, application developers typically use the default parameters and often experience poor I/O performance. This work conducts a benchmarking-based analysis on the HDF5 behaviors with a wide variety of I/O patterns to extract performance characteristics under the production workload. To make the analysis well controlled, we exercise I/O benchmarks on POSIX-IO, MPI-IO, and HDF5 using the same I/O patterns and in the same jobs. To address high performance variability in production environments, we repeat the benchmarks across I/O patterns, storage devices, and time intervals. Based on the results, we identified consistent HDF5 behaviors that appropriate configurations and operations on dataset layout and file-metadata placement can improve performance significantly. We apply our findings and evaluate the tuned I/O library on two supercomputers: Summit and Cori. The results show that our tuned parameters can achieve more than 10× I/O performance speedup than that with default parameters on both systems, suggesting the effectiveness, stability, and generality of our solution.

Xie, Bing↗
NASA NTRS2002

NAS Parallel Benchmarks I/O Version 2.4

We describe a benchmark problem, based on the Block-Tridiagonal (BT) problem of the NAS Parallel Benchmarks (NPB), which is used to test the output capabilities of high-performance computing systems, especially parallel systems. We also present a source code implementation of the benchmark, called NPBIO2.4-MPI, based on the MPI implementation of NPB, using a variety of ways to write the computed solutions to file.

Wong, Parkson↗
DOE OSTI2020

Striped Data Analysis Framework

A columnar data representation is known to be an efficient way for data storage, specifically in cases when the analysis is often done based only on a small fragment of the available data structures. A data representation like Apache Parquet is a step forward from a columnar representation, which splits data horizontally to allow for easy parallelization of data analysis. Based on the general idea of columnar data storage, working on the [LDRD Project], we have developed a striped data representation, which, we believe, is better suited to the needs of High Energy Physics data analysis. A traditional columnar approach allows for efficient data analysis of complex structures. While keeping all the benefits of columnar data representations, the striped mechanism goes further by enabling easy parallelization of computations without requiring special hardware. We will present an implementation and some performance characteristics of such a data representation mechanism using a distributed no-SQL database or a local file system, unified under the same API and data representation model. The representation is efficient and at the same time simple so that it allows for a common data model and APIs for wide range of underlying storage mechanisms such as distributed no-SQL databases and local file systems. Striped storage adopts Numpy arrays as its basic data representation format, which makes it easy and efficient to use in Python applications. The Striped Data Server is a web service, which allows to hide the server implementation details from the end user, easily exposes data to WAN users, and allows to utilize well known and developed data caching solutions to further increase data access efficiency. We are considering the Striped Data Server as the core of an enterprise scale data analysis platform for High Energy Physics and similar areas of data processing. We have been testing this architecture with a 2TB dataset from a CMS dark matter search and plan to expand it to multiple 100 TB or even PB scale. We will present the striped format, Striped Data Server architecture and performance test results.

Gutsche, Oliver↗
DOE OSTI2018

Striped Data Server for Scalable Parallel Data Analysis

A columnar data representation is known to be an efficient way for data storage, specifically in cases when the analysis is often done based only on a small fragment of the available data structures. A data representation like Apache Parquet is a step forward from a columnar representation, which splits data horizontally to allow for easy parallelization of data analysis. Based on the general idea of columnar data storage, working on the [LDRD Project], we have developed a striped data representation, which, we believe, is better suited to the needs of High Energy Physics data analysis. A traditional columnar approach allows for efficient data analysis of complex structures. While keeping all the benefits of columnar data representations, the striped mechanism goes further by enabling easy parallelization of computations without requiring special hardware. We will present an implementation and some performance characteristics of such a data representation mechanism using a distributed no-SQL database or a local file system, unified under the same API and data representation model. The representation is efficient and at the same time simple so that it allows for a common data model and APIs for wide range of underlying storage mechanisms such as distributed no-SQL databases and local file systems. Striped storage adopts Numpy arrays as its basic data representation format, which makes it easy and efficient to use in Python applications. The Striped Data Server is a web service, which allows to hide the server implementation details from the end user, easily exposes data to WAN users, and allows to utilize well known and developed data caching solutions to further increase data access efficiency. We are considering the Striped Data Server as the core of an enterprise scale data analysis platform for High Energy Physics and similar areas of data processing. As a result, we have been testing this architecture with a 2TB dataset from a CMS dark matter search and plan to expand it to multiple 100 TB or even PB scale. We will present the striped format, Striped Data Server architecture and performance test results.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗
NASA NTRS1999

I/O Parallelization for the Goddard Earth Observing System Data Assimilation System (GEOS DAS)

The National Aeronautics and Space Administration (NASA) Data Assimilation Office (DAO) at the Goddard Space Flight Center (GSFC) has developed the GEOS DAS, a data assimilation system that provides production support for NASA missions and will support NASA's Earth Observing System (EOS) in the coming years. The DAO's support of the EOS project along with the requirement of producing long-term reanalysis datasets with an unvarying system levy a large I/O burden on the future system. The DAO has been involved in prototyping parallel implementations of the GEOS DAS for a number of years and is now converting the production version from shared-memory parallelism to distributed-memory parallelism using the portable Message-Passing Interface (MPI). If the MPI-based GEOS DAS is to meet these production requirements, we must make I/O from the parallel system efficient. We have designed a scheme that allows efficient I/O processing while retaining portability, reducing the need for post-processing, and producing data formats that are required by our users, both internal and external. The first phase of the GEOS DAS Parallel I/O System (GPIOS) will expand upon the common method of gathering global data to a Single PE for output. Instead of using a PE also tasked with primary computation, a number of PEs will be dedicated to I/O and its related tasks. This allows the data transformations and formatting required prior to output to take place asynchronously with respect to the GEOS DAS assimilation cycle, improving performance and generating output data sets in a format convenient for our users. I/O PEs can be added as needed to handle larger data volumes or to meet user file specifications. We will show I/O performance results from a prototype MPI GCM integrated with GPIOS. Phase two of GPIOS development will examine ways of integrating new software technologies to further improve performance and build scalability into the system. The maturing of MPI-IO implementations and other supporting libraries such as parallel HDF should provide performance gains while retaining portability.

Lucchesi, R.↗
DOE OSTI2025

HDF5 in the exascale era: Delivering efficient and scalable parallel I/O for exascale applications

Accurately modeling real-world systems requires scientific applications at exascale to generate massive amounts of data and manage data storage efficiently. However, parallel input and output (I/O) faces challenges due to new application workflows and the state-of-the-art memory, interconnect, and storage architectures considered in exascale designs. The storage hierarchy has expanded with node-local persistent memory, solid-state storage, and traditional disk and tape-based storage, thus requiring efficiency at each layer and much more efficient data movement among these layers. This paper discusses how the ExaHDF5 project improved the I/O performance and data management for exascale architectures by enhancing HDF5, a widely used parallel I/O library. The team developed an Asynchronous I/O Virtual Object Layer (VOL) connector that allowed overlapping I/O with computation. They also created a Cache VOL to complement asynchronous I/O by incorporating fast storage layers, such as burst buffer and node-local storage, into the parallel I/O workflow through caching and staging data. Additionally, the team enabled data aggregation and I/O at the node level by using a Subfiling Virtual File Driver (VFD). To demonstrate superior I/O performance with HDF5 at exascale, the ExaHDF5 team collaborated with several exascale applications. In this paper, we show I/O performance improvements for three applications: Cabana (a particle-based simulation library), EQSIM (a regional earthquake simulation software), and E3SM (a climate system modeling library).

Asynchronous I/Ol↗