Service status
Current availability of the commons, by region and by service. Figures cover the 90 days to 4 November 2026.
- Served today
- 61.4 TB
- Active transfers
- 3,812
- Median start time
- 0.31 s
Regions
| Region | Location | State | 90-day uptime | p50 latency | Stored |
|---|---|---|---|---|---|
| nm-eu-north | Nordic data hall, member-hosted | Operational | 99.97% | 41 ms | 26 PB |
| nm-eu-central | Frankfurt colocation | Operational | 99.96% | 29 ms | 26 PB |
| nm-us-central | US Midwest colocation | Operational | 99.93% | 46 ms | 24 PB |
| nm-ap-south | Singapore colocation | Operational | 99.90% | 58 ms | 19 PB |
Latency is the median time to first byte measured from probes in 22 networks. Stored capacity is raw usable capacity, not unique data; data is replicated across at least three regions.
Where traffic goes
How routing works
Requests are answered by the nearest healthy region. A region is marked unhealthy when its error rate stays above 2 percent for three minutes, and traffic is drained in steps of 25 percent. Signed manifests list at least two regions, so clients such as aria2 continue on the next mirror without intervention.
Availability targets are 99.5 percent for Associate members and 99.9 percent for Full members, measured monthly on HTTPS delivery. Credits are described in the membership agreement.
Services, last 90 days
Catalogue and landing pagesStatic and cached
HTTPS file deliveryRanged reads, signed URLs
Manifest servicearia2 and rclone manifests
Transfer endpointsGlobus-style, members only
Account consoleSign-in, tokens, billing
Each bar is one day, oldest on the left. Amber bars mark days with a degradation of 10 minutes or more.
Incident history, 2026
Elevated errors in nm-ap-south Resolved
A storage controller replacement left 6 percent of requests in Singapore returning 502 for 38 minutes. Traffic was shifted to nm-eu-central and the controller was reinstated after verification.
Manifest generation delayed Resolved
A schema migration slowed signed-manifest generation for large collections. Requests for GEN-0388 and SAT-2031 manifests took up to 14 minutes for 72 minutes. No data was affected.
Planned maintenance in nm-us-central Completed
Network equipment upgrade, 3 hours inside the announced 6 hour window. Requests were served from other regions; median latency in the Americas rose by 52 ms.
Signature key rotation caused expired_signature errors Resolved
A new signing key reached edge nodes 22 minutes before the old one was retired. Clients with manifests older than a day received HTTP 403 until caches were purged. Rotation now overlaps for 48 hours.
Packet loss on a transit link to nm-eu-north Resolved
A transit provider fault caused 2 to 4 percent loss for 96 minutes. Throughput for Nordic clients fell by about a third; other regions were unaffected.