What happened

AWS reported packet loss affecting Direct Connect connections at Equinix FR5 in Frankfurt, Germany. Its final update attributes the incident to water entering the facility, affecting cooling and power. Customers with redundant connections at other Direct Connect locations maintained connectivity through alternate paths.

Timeline (UTC)

TimeProvider-reported milestone
August 15, 02:33Impact began
August 15, 03:42First public notice
August 18, 02:26Equipment restored; connectivity verified operational
August 18, 04:01Final resolved update published

The start and recovery times are converted from AWS's stated Pacific Daylight Time. Their interval is 71 hours, 53 minutes, not a measurement of every customer's downtime. Recovery occurred in stages.

Source and scope

See AWS Health and its public historical event record, under directconnect-eu-central-1, event ending E23AA_ADAB9D3D11F. This record describes a specific Direct Connect location; it does not establish an AWS-wide outage.

Correction and citation

Corrected September 12, 2026: the previous 96-hour, 20-minute duration and August 19 recovery were unsupported. This revision uses AWS's final incident update. Cite this brief with its location, service scope and UTC times intact.

For present conditions, check current AWS status.