This page documents the inaugural meeting of the Array Database assessment Working Group at Plenary 8 in Denver, USA, 2016-09-16.
In order to maximise focus in getting hands-on, just two introductory presetnations were given by the chairs, then leading into a discussion on how to proceed concretely.
- Q: what is the focus of this WG?
A: Focus is strictly on arrays (grids, rasters) to be able to achieve meaningful results
- Q: Are only array databases in focus?
A: Anything that offers a flexible query interface is of interest, and could be called a "database system" in the widest sense. However, further array-related technology should be inspected as well to obtain comparisons (e.g., MapReduce-based systems like SciHadoop, GeoTrellis, MrGeo) - but this depends on the contributions; in other words: come and champion your favourite tool!
- Q: what is a realistically large deployment?
A: anything larger than server RAM qualifies (otherwise, you just load data into RAM without any need for a database system); as a rule of thumb, going beyond 10 TB should be fine.
- Q: what kind of contributions do you seek?
A: three types of contributors are needed in order to have comprehensive expertise on board:
- use case providers: they bring in data and challenges (aka queries)
- service providers: they provide storage and compute space for setting up realistically large deployments of systems
- technology (software) providers: they provide their tools for installation and operation, as well as support.
- In an elicitation of the interests of those present the following was articulated:
- ETL: 2
- SQL/MDA: 4
- formats (HDF, ...): 5
- query power (Fourier Transform etc.): 3
- OPeNDAP: 2
- model-to-model communication via database: 2
The following concrete initial groups shaped:
- use case providers:
- Mohan: Earth science
- Schaaff: Space science
- data center: none
- software providers:
- Peter Baumann: rasdaman
- Ted Habermann: HDF
- Mohan: HDF, OPeNDAP
- Doug Lindholm: tbd
- chairs: convene group, grow it from membership via email - until beginning of Oct
- data center group: establish infrastructure for installation - tbd
- software providers: stand up installations, provide tutorials - once installation points are defined