This software repository contains source codes implementing Log VOL connector, an HDF5 Virtual Object Layer (VOL) plugin that stores HDF5 datasets in a storage layout similar to the time log. When using the Log VOL connector, write requests from an MPI process are appended one after another (as logs) in a contiguous space in the file. The contiguous spaces of multiple processes are appended one after another following the increasing order of processes' MPI rank IDs. Such a log layout I/O strategy avoids the expensive inter-process communication required when storing data in the canonical order. One of the pwoerful features of HDF5 VOL is to allow applications to make use a VOL connector by setting two environment variables without changing the application source codes. Using the Log VOL connector, exist HDF5 programs can achieve a better parallel write performance with no changes to their codes. Files created by the Log VOL conform with the HDF5 file format specification, but require the Log VOL to read them back.
- Current build status:
- This Log VOL connector has been registered with the HDF group with Connector Identifier 514.
- doc/userguide.md contains the compile and run instructions.
- doc/design.md outlines the design of Log VOL connector.
- doc/api.md describes the new APIs introduced in this VOL.
- E3SM I/O case study - Energy Exascale Earth System Model (E3SM).
- WRF I/O case study - Weather Research and Forecasting Model (WRF).
- Wei-keng Liao <[email protected]>
- Kai-yuan Hou <[email protected]>
- Zanhua Huang <[email protected]>
Copyright (C) 2022, Northwestern University. See COPYRIGHT notice in top-level directory.
Ongoing development and maintenance of the Log VOL connector are supported by the Exascale Computing Project (17-SC-20-SC), a joint project of the U.S. Department of Energy's Office of Science and National Nuclear Security Administration, responsible for delivering a capable exascale ecosystem, including software, applications, and hardware technology, to support the nation's exascale computing imperative.