Limits and quotas
Limits are published so you can design around them. Where a limit can be raised, the last column says how.
Objects
| Limit | Starter | Team | Scale | Enterprise |
|---|---|---|---|---|
| Largest container layer | 5 GB | 50 GB | 100 GB | 200 GB |
| Largest generic object | 5 GB | 50 GB | 200 GB | 1 TB |
| Largest package file | 500 MB | 2 GB | 5 GB | 10 GB |
| Tags per repository | 1,000 | 100,000 | 1,000,000 | No fixed limit |
| Largest cache entry | 5 GB | 40 GB | 40 GB | 200 GB |
Request rates
| Operation | Starter | Team | Scale | Notes |
|---|---|---|---|---|
| Blob and package reads | 6,000 per day | No limit | No limit | Bandwidth is limited per organisation to 40 Gbit/s on Team and 200 Gbit/s on Scale per region. |
| Manifest and metadata reads | 60 per minute | 3,000 per minute | 12,000 per minute | Per token. A 429 includes a Retry-After header. |
| Writes | 30 per minute | 600 per minute | 3,000 per minute | Counted per manifest or version, not per blob. |
| Management API | 120 per minute | 600 per minute | 1,800 per minute | Per organisation. |
Retention and recovery
- Soft-deleted artifacts can be restored for 7 days on all plans.
- Audit log retention is 7 days on Starter, 90 days on Team, 1 year on Scale.
- Cache entries unused for 14 days are removed by default; the setting is per organisation.
- Mirrors revalidate tags after 6 hours and package metadata after 30 days.
Replication and regions
Team organisations may replicate to one additional region and Scale organisations to three. Replication is asynchronous, with a median lag of 1.4 seconds for 100 MB between eu-north-1 and eu-central-1 and 3.1 seconds between eu-north-1 and us-east-1. Enterprise organisations can use a dedicated storage cell in any region, with synchronous replication between two availability zones.
Need a higher limit on a paid plan? Write to support@fi4.enotcloud.com with the organisation slug and the use case. Most object size increases are approved the same day.