Skip to content

Data Studios deep dive

This content is transcribed from a Data Studios demo presented by Rob Newman.

  1. Create a custom data link:

    • Use the Data Explorer to add a specific data directory.
    • Select Add cloud bucket and specify the exact path to your data.
    • Note: Any data link added to a data studio session is read/write.

  2. Directory isolation:

    • Once a directory is mounted to a data studio session, it cannot be accessed outside of that session.
    • This isolation prevents others from overwriting your results.
    • This is important ensure that only designated scientists or bioinformaticians can work with specific project directories.

  3. Fusion symlinks limitation:

    • Fusion symlinks will not work outside the specified directory.

  4. Allowed buckets:

    • Ensure the buckets you want to access through Data Studios are listed in the Allowed buckets section in the compute environment (CE).

Compute environment resources

  1. Resource management:

    • When you use a compute environment (CE) for both pipelines and Data Studios sessions, they will compete for resources.
    • To avoid stalling or losing work due to lack of CPU or memory, consider using a separate CE for Data Studios sessions.

  2. Large file handling:

    • Staging large files (such as BAM files) can crash your session if there aren't enough resources.
    • Ensure your data studio has adequate resources before working with large datasets.

  3. Identifying data studio sessions:

    • Each studio session has a name that can be identified in AWS Batch within the appropriate CE as the running job name.

Data Studios infrastructure and mechanics

  1. Container Web Server:

    • Each Data Studio container includes a web server called Tower Connect, which communicates telemetry data to and from Platform.
    • Custom containers may add a layer that includes this web server.

  2. Snapshots:

    • Snapshots are created when you first create a session and each time you stop a studio.
    • These snapshots help in saving and restoring your work environment.

Collaboration

  1. Monitoring activity:

    • If a user is inactive for more than 5 minutes, their icon will disappear from the monitoring page, indicating they are not currently collaborating.

  2. Real-time collaboration:

    • Only Jupyter and VS Code support real-time collaboration.
    • RStudio requires a Pro license for real-time collaboration — discussions with Posit for this capability are ongoing.