Amazon SageMaker Studio Space
Does ZopNight manage Amazon SageMaker Studio Space?
SageMaker Studio spaces bill through storage, not compute: each space keeps an EBS-backed volume that charges per GB-month whether or not anyone opens the space, while the apps launched from it meter separately per instance-hour. ZopNight discovers spaces on the 6-hour cycle, tracks their storage cost, and flags abandoned spaces.
Rules that fire on Amazon SageMaker Studio Space
No active rule family targets Amazon SageMaker Studio Space today. Rules that used to are retired, and retired rules publish no pages and fire no findings. Scheduling and permissions coverage are unaffected.
At a glance
| Field | Value |
|---|---|
| Scheduling notes | discovery and cost tracking only. |
A SageMaker Studio space is a persistent workspace holding storage and app configuration for Studio users. Spaces carry EBS-backed storage cost and anchor the apps whose compute bills accumulate.
The workspace that keeps its disk
A space is Studio’s unit of persistence: the files, environment configuration, and app definitions that survive between sessions. Its own meter is storage: the EBS-backed volume behind each space bills per GB-month continuously, opened or not. Compute is adjacent rather than intrinsic: apps launched within the space bill their own instance-hours and stop billing when stopped, but the space’s volume persists through it all. That makes spaces the ML equivalent of home directories on a shared server, individually cheap, collectively unbounded, and never cleaned because they might contain something.
Abandonment detection for workspaces
ZopNight discovers spaces automatically on the 6-hour cycle, with storage cost from Cost Explorer or CUR 2.0 and hygiene recommendations for abandoned spaces. The abandonment signal is usage-shaped: a space whose apps have not run in months is a workspace whose owner has moved on (to another team, another tool, or another employer) while its volume bills monthly. Spaces also matter as anchors: an abandoned space sometimes still holds a running app, and the space-level view is where that pairing becomes visible.
How workspace storage compounds
Data gravity does the work. ML workspaces accumulate datasets copied “temporarily” for an experiment, model checkpoints from every training attempt, and conda environments duplicated per project. Volumes grow to hold them, and volume growth is one-directional in practice. Per-user-per-project space conventions multiply the baseline: an eight-person team with three projects runs two dozen volumes. Offboarding rarely includes Studio cleanup, so departed users’ spaces persist as archaeology.
Space inventory per domain
The SageMaker console’s Studio section lists spaces per domain with their storage settings and last-modified activity. The audit pairs each space with a present-tense owner; spaces that cannot produce one get their volumes snapshotted if anyone insists, and deleted either way.