IBM Watson Data & AI - Structured Ideas

This Ideas portal is being closed.  Please enter new idea at

Add the ability to add/remove OS-level libs for an entire cluster.

Currently, it is not possible to install new OS-level libraries in DSX, due to a lack of sudo permissions. Unfortunately, this prevents DSX from being a viable environment for many actual machine learning projects, such as those that require reading specialized data formats for which a specialized OS-level library is needed.

  • Mike Dusenberry
  • Dec 14 2018
  • Needs review
Customer Name
Role Summary
  • Attach files
  • Roland Weber commented
    December 14, 2018 17:59

    There are now different backends you can use. With IBM Analytics Engine, you get your own cluster and can configure it to your needs. With Environments, you get a single-machine container and no sudo access, but libraries available from conda can be installed. The original Spark as a Service cluster is shared, so users cannot be allowed to make cluster-wide changes there.