Cloud & infrastructure status
Providers whose outages take hundreds of unrelated sites with them. All 9 are checked every 20 minutes; the status beside each one is the result of its most recent check.
Cloud infrastructure is the only category here where one outage takes down hundreds of unrelated companies at once, and where most people never learn the real cause of a site they could not reach.
How reliable cloud & infrastructure has been, last 30 days
Measured by our own checks, run every 20 minutes, not taken from any provider’s status page. 9 of 9 services have enough history to report.
| Service | Now | Uptime 30d | Downtime | Response |
|---|---|---|---|---|
| Amazon Web Services | Up | 100.0% | none | 182 ms |
| Cloudflare | Up | 100.0% | none | 410 ms |
| Cloudflare DNS (1.1.1.1) | Up | 100.0% | none | 163 ms |
| DigitalOcean | Up | 100.0% | none | 122 ms |
| Dropbox | Up | 100.0% | none | 360 ms |
| Google Drive | Up | 100.0% | none | 440 ms |
| Google Public DNS (8.8.8.8) | Up | 100.0% | none | 163 ms |
| iCloud | Up | 100.0% | none | 150 ms |
| Ring | Up | 100.0% | none | 198 ms |
DigitalOcean has been the steadiest here over the last 30 days. Uptime is observed uptime: The share of measured time our checks got a usable answer. It excludes nothing, including announced maintenance.
Why cloud & infrastructure services go down
A handful of providers host a very large share of the web, concentrated further into a few enormous regions. When one region degrades, the failure surfaces as dozens of separate consumer brands breaking simultaneously, none of which mention the provider. This is the single most important thing to understand about modern outages: The site you could not reach is often not the thing that failed.
The dependency is also deeper than hosting. DNS, identity, content delivery and payment processing are each concentrated in a few providers, and services depend on each other in ways their own engineers have not fully mapped. A dependency failure produces the widest blast radius of anything measured on this site.
What an outage here looks like
Many unrelated services fail at once, in the same minutes, for the same people. Errors are inconsistent - some sites time out, others return a gateway error, others load partially - because each one fails at whatever layer it depended on.
When these services are under most strain
No daily pattern. Cloud incidents correlate with change: Deploys, capacity events and configuration pushes, which is why a quiet Tuesday afternoon is as likely as any other time.
At a glance
- Typical duration
- 1 to 6 hours, with a long tail as dependents recover
- Usual cause
- Regional capacity, configuration push, or DNS
- Tell-tale sign
- Several unrelated brands failing in the same minute
- Blast radius
- The widest of any category here
What to do when cloud & infrastructure is down
- Notice the pattern before you troubleshoot. Three unrelated sites failing together is not three coincidences. Stop checking your own equipment.
- Read the provider status page, not the brand one. The brand is usually as much in the dark as you are, and their status page will update last.
- There is nothing to do but wait. This is the one category where no user-side action helps at all. Retrying puts load on a recovering system.
Still not sure whether it is them or you? Check the site from our servers - We are on a different network, a different resolver and a different route from you.