Question

Your data science team uses Vertex AI Workbench managed notebooks for model prototyping. Several data scientists frequently forget to stop their instances after finishing exploratory work, leading to significant idle GPU costs. Leadership asks you to reduce these costs without disrupting active development sessions. What should you configure?

Accepted Answer

Enable idle shutdown on the managed notebook instances so they automatically stop after a configured period of inactivity.

Answer

Set up a Cloud Scheduler job that force-deletes all notebook instances every night at midnight.

Answer

Migrate all prototyping work to Colab Enterprise and remove Workbench access entirely.

Answer

Replace the GPU instances with CPU-only instances to eliminate GPU charges.