It tends to be more library dependencies than live clusters. A lot of data lakes are managed using Hadoop and Spark so I think it’s just an artefact of that.

It tends to be more library dependencies than live clusters. A lot of data lakes are managed using Hadoop and Spark so I think it’s just an artefact of that. In the end I can’t see why you wouldn’t just be able to create and manage Iceberg files directly from a standard Python/JS/Java without that legacy.