...
Topic | Notes |
---|---|
SDAP Plugins / Plugable algorithms | |
Discussion: Zarr dataset performance issues with Spark | Accessing Zarr data works fine, but opening the dataset can cause excessive delays, especially depending on how the time coordinate is chunked. This is worse for Spark endpoints as all the workers need to do this and it isn't guaranteed to remain open between queries. Need advice for a workaround. |