Storage Guide

Storage Guide#

Quick Reference: Cheat Sheet#

  • MyData → Your personal notebook

    • Private, backed up, long-term

    • Best for personal files, scripts, tests

  • MyProjects → Shared filing cabinet

    • Collaborative, long-term, curation-ready

    • Best for team data, archiving, publication

  • Work → Scratchpad for active jobs

    • High-performance, temporary, not backed up

    • Best for staging inputs & writing job outputs

Rule of Thumb:

  • If it’s important, back it up (MyData or MyProjects).

  • If it’s temporary, run it in Work.

Storage Decision Tree#

Which Storage Should I Use?

  • Step 1: Are you running a compute job?

    • Yes → Use Work for job inputs/outputs (fast I/O).

    • No → Continue ↓

  • Step 2: Do you need to share the data with a team?

    • Yes → Use MyProjects (collaboration, publication).

    • No → Continue ↓

  • Step 3: Do you just need a private workspace?

    • Yes → Use MyData (personal, backed up).

    • No → Continue ↓

  • Step 4: Is this public reference data you need to read?

    • Yes → Access from CommunityData or Published (read-only).

    • No → Re-check your workflow; you may be mixing storage types.

In short:

  • Work = jobs (fast, temporary).

  • MyProjects = team (shared, curation-ready).

  • MyData = personal (private, backed up).

  • Community/Published = reference only.

Common Mistakes to Avoid#

Don’t leave important results in Work — it’s not backed up

  • Work is not backed up and may be purged without warning.

  • Always copy valuable results to MyProjects or MyData.

Don’t use MyData for team projects — use MyProjects instead

  • Files in MyData are private unless explicitly shared.

  • Use MyProjects for collaboration.

Don’t try to save to Community/Published — they’re read-only

  • These directories are read-only.

  • They are for reference only.

Don’t run jobs directly from Corral — stage to Work first

  • Corral is network-mounted and slower for I/O-intensive workloads.

  • Always stage input/output to Work for active jobs.

Don’t Use Work for long-term project storage

  • Work is high-performance but temporary.

  • Use MyProjects for durable project data.

Remember:

  • Corral = archive

  • Work = scratch

  • Node-local = temporary runtime.

Contents