Skip to content

Investigate using MariaDB as storage for pipeline's working set #21

@SteveOv

Description

@SteveOv

Likely to be more memory efficient as the client need only hold and update one target's data at any time. Also more durable, as a failure will only affect one target.

This will allow the pipeline to scale out, with potentially multiple client instances running

The current approach of using a QTable does now work but it relies on reading in the whole file, working on all selected targets before overwriting the file with an updated copy at the end of the step. If anything goes wrong, we get none of the updates.

Metadata

Metadata

Assignees

Labels

enhancementNew feature or requestquestionFurther information is requested

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions