Microsoft Azure
Cloud infrastructure · Official status page (opens in a new tab)
- 3Failures in Microsoft Azure's systems
- 0Its services disrupted by failures elsewhere
- Jan 25, 2023latest Oct 29, 2025Documented since
- 3 of 3Own failures with root cause confirmed
Counts cover incidents in this archive only. They are not a complete outage history or a reliability measure.
Failures in Microsoft Azure's systems
Incidents where the documented failure was in Microsoft Azure's own systems. Durations are measured to the end points the sources give.
- Oct 29, 2025Azure Front Door outage: a valid customer change exposes a latent bug at the edgeDuration: 8 h 24 minto: Impact mitigatedResolvedAzure Front DoorAzure CDNAzure Portal+6Software defect · Root cause confirmed
Stated cause
A valid customer configuration change produced incompatible metadata that "exposed a latent bug in the data plane" during asynchronous processing. Edge servers crashed, which also affected Front Door's internal DNS service. Microsoft also cites a gap in pre-production validation.
- Jul 30, 2024Azure Front Door outage: a DDoS attack exposes a network misconfigurationDuration: 2 h 13 minto: Most impact mitigatedResolvedAzure Front DoorAzure CDNAzure Portal+2Configuration change · Root cause confirmed
Stated cause
Microsoft says a DDoS attack "was merely a trigger event." The underlying cause was that "a network misconfiguration caused congestion and packet loss for AFD frontends," starting in Europe and spreading across regions.
- Jan 25, 2023Microsoft cloud outage: one router command disrupts Azure's global networkDuration: 5 h 35 minto: Full mitigationResolvedAzure regionsMicrosoft 365Power Platform+1Human error · Root cause confirmed
Stated cause
While adding WAN capacity in Madrid, an engineer ran a standard operating procedure that had been modified without retesting. It included a command to purge the IGP database which, on one vendor's routers, runs across all routers joined to the IGP. All routers recomputed connectivity and BGP routers re-advertised Internet prefixes. The same operation ran on a second Madrid router 33 minutes later.