How to download large amount of data from Teams files

For 6–7 TB of Teams/SharePoint content, browser-based “Download” and ad‑hoc sync are not suitable because of size and ZIP limits, and because placeholders do not hydrate automatically in a way that supports tools like Robocopy.

From the provided information, the supported large‑scale pattern is:

1. Use a migration tool or service rather than browser download
   • Browser download from OneDrive/SharePoint creates a ZIP and is subject to limits: 250 GB per file, 10,000 files per download, and 10 GB per ZIP file with a 20 GB overall compressed limit for multiple compressed files. Large folders can therefore fail or produce incomplete ZIPs.
   • This aligns with the behavior described (downloads failing or ZIPs missing content).
2. Avoid Azure Data Box unless the bottleneck is connectivity
   • Azure Data Box is intended when the primary issue is poor connectivity or constrained source resources, not just “a large amount of data.”
   • It adds complexity: data must first be copied to the Data Box, shipped, ingested into Azure, and then the same migration steps must be run again from Azure to the final target.
   • Most large enterprise customers (including those with >500 TB) achieve migration using multiple computers and tools such as Migration Manager rather than Data Box.
3. Respect the documented download limits and behavior
   • When selecting multiple files/folders and choosing Download in OneDrive/SharePoint, the service creates a ZIP with all selected content, but:
     • Total number of files is limited to 10,000.
     • Folder size is limited to 10,000 files.
     • Compressed downloads have a 10 GB per ZIP and 20 GB overall limit for multiple compressed files.
     • Files encrypted by sensitivity labels or IRM over 4 MB are excluded from the ZIP and must be downloaded separately.
   • These limits explain why large folder downloads from the linked SharePoint site are unreliable for multi‑TB migrations.
4. Use a file‑level approach that preserves structure
   • To maintain folder structure to a NAS, the migration should operate on individual files and folders rather than ZIP archives.
   • The documented guidance for large migrations to Microsoft 365 is to use multiple computers and a migration tool (for example, Migration Manager) instead of Data Box. In reverse (from SharePoint/Teams down to on‑prem/NAS), a similar principle applies: rely on a tool or process that enumerates and copies files directly, honoring the hierarchy and avoiding ZIP size limits.
5. Why sync + Robocopy from the cache fails
   • The Cloud Files API design expects the sync provider to own hydration of placeholders, not File Explorer or tools like Robocopy. Copying large directories with multi‑GB placeholders by relying only on implicit hydration is not reliable.
   • This matches the behavior where placeholders appear locally but cannot be robustly copied with Robocopy/Explorer from the cache folder.

Given the constraints in the provided material, the supported guidance is:

• Do not rely on browser ZIP downloads for multi‑TB Teams/SharePoint content because of documented limits and failure modes.
• Do not choose Azure Data Box solely because the dataset is large; it is recommended only when connectivity or source resource constraints are the true bottleneck.
• Use a migration approach that:
  • Enumerates and copies files directly (not as a single ZIP),
  • Runs from multiple machines in parallel if needed,
  • Ensures files are hydrated before copy when using sync technologies.

This approach preserves folder structure and avoids the ZIP and placeholder limitations described in the documentation.

References:

• Download files and folders from OneDrive or SharePoint
• Migrate your file share content to Microsoft 365 using the Azure Data Box
• Cloud File API FAQ-2 - Microsoft Q&A