Next-generation Data Model Reference Implement not ready for production use

[hjedwards]

Description

The OpenCDMS project can be relatively successful by improving interoperability among existing solutions. However, the long-term aim is to deliver a solution with full support for the requirements in the WMO Climate Data Management System Specifications (WMO 2014) and to create a next-generation solution that goes beyond the specifications.

Mitigation

OpenCDMS will adopt on of the data models from an existing supported system as the default data model in the absence of the reference implementation data model.

[Steve-Palmer]

This is an important decision to take forward the project. The choices are to either

1. Develop a data model from first principles, and then adapt the rest of the project to use this, OR
2. Adopt an existing data model which is known to work for many (but not all) use cases, and then adapt and extend this in the light of experience with additional data types and use cases.

In either case, there are important decisions to be made on the scope of the data model, especially for the scope of the metadata. In particular, there have been moves to include more historical and calibration information in the recorded metadata rather than assuming that these metadata items are held in other systems (which may or may not have been archived and made available).

I would support option 2 - adoption of an existing data model - as a practical way to implement trial systems, and then learn and adapt. The risks of using option 1 - design from first principles - are that development and learning are delayed, and that consensus on the first-principles can be obtained without a demonstration system.