Skip to end of metadata
Go to start of metadata


Meeting Info

1.  Please join my meeting.

2.  Use your microphone and speakers (VoIP) - a headset is recommended. Or, call in using your telephone.

Dial +1 (224) 501-3412
Access Code: 840-709-413
Audio PIN: Shown after joining the meeting

Meeting ID: 840-709-413



  • Go over agenda items

Discussion items

5 minUpdates to CharterElenaHDF5 Technical Advisory Board Charter
10 minPrabhat's feedback on strengths and weaknessesAllHDF5 Software Strengths and Weaknesses
40 min2019 Plan 

From Charter: At the beginning of the calendar year the Board will identify goals and objectives to be achieved during the year. Those will be summarized in a document that will be available to the HDF5 community. By the end of the year the Board will publish a report on its activities and achievements.

Emerged motives from the previous discussions of strengths and weaknesses:

  • A set of large-scale use cases to understand performance issues and develop necessary improvements and tactics one of which might be log structure that would bring HDF5 up to par with these other libraries (ADIOS).
      • Elena will look at the paper and coordinate with the existing projects; report at the next TAB meeting.
    • How well benchmarks are accepted by large community? Maybe the first thing is to evaluate and if they are appropriate for. Build DB of performance results.
    • If the goal is how to use correctly then this is not the right approach.
    • Benchmarks are baseline for performance on different systems, understanding the performance and having a roadmap to fix it.
    • Not use word "benchmark".
  • Discussion of different motives/examples users can chose from to avoid misuse of HDF5 library.
    • Approach this one first and then come back to the first one.
    • Have several examples - particle example, etc,. intentions of I/O operations are, and codes to demonstrate how to do those well.

Identify examples to cover for this year, which examples make sense to cover. Shared file I/O, multiple files I/O - two different approaches to cover.

Performance of HDF5 on top of DAOS. There were several synthetic codes  - legion, climate code. We need a balance of the set of use cases looks like.

Another dimension: do we want the use cases to focus on callers of HDF5 directly, or something like CGNS and netCDF-4. Do they use HDF5 correctly or there are weaknesses in HDF5?

Functionality for both reads and writes; search - data and metadata, query interfaces.

Capability should be performant.

Examples should be done at the workflow level. Emphasize schema and constraints.


  • Come up with the use case example and a description of schema, storage
  • Put on Confluence.

Action items