EP R&D Software Working Group Meeting

Wednesday 10 Feb 2021, 10:00 → 11:00 Europe/Zurich

Vidyo

Live Notes

Software R&D Working Meeting Minutes, 2021-02-10
====

Plans for 2021
---

Will cover the tracking tasks next time around.

To organise/Doodle for the next meetings, up to the summer.

Efficient Analysis
---

There is a possibility to do also lossy compression, e.g., zeroing out
high precision bits (which are just noise) to improve compression.
(Although this is not currenly in the R&D PoW.)

Talk to the experiments, as a lot of work was done to try reduce the
number of bits written, e.g. in the tracking case reducing a lot the
information needed to store. (Andi)

Not clear that everything should go in the I/O layer (almost certainly not),
but the I/O layer should be flexible enough to work with lossily
compressed data.

Do object stores have consistent conceptual laters?
- No, very a lot in what an object actually maps too... (CEPH, S3, etc.)
    - Could choose a page or some kind of super-cluster
- May well need different kinds of layouts for different stores
- DAOS, v1 release, there is an active R&D programme
    - No formal standardisation that we know of
    - Targets HPC market, rather than general data centre storage
        - Relies heavily on Intel's (expensive) hardware

10:00 → 10:15 Introduction and Plans for 2021

Speakers: Graeme A Stewart (CERN), Jakob Blomer (CERN)

WP7 Working Meeting Intro 2021-02-10.pdf

10:15 → 10:45 Efficient Analysis Systems

Speakers: Enric Tejedor Saavedra (CERN), Jakob Blomer (CERN), Javier Lopez Gomez (CERN), Vincenzo Eduardo Padulano (Valencia Polytechnic University (ES))

EP R&D Software - Efficient Analysis Facilities (2nd Update).pdf

10:45 → 11:00 Roundtable and Discussion