At approximately 10:00 AM EDT on September 2, Cogent experienced a large-scale backbone fiber incident affecting Los Angeles and the Southwestern United States. All LAX-BGP VPS services were affected to varying degrees.
Cogent confirmed multiple backbone fiber cuts, including incidents in Modesto and between El Centro and Calipatria, California. This incident originated within Cogent's network and fiber infrastructure and was not caused by customer VPS instances, host servers, or NovixLink equipment.
UPDATE | September 3, 2:10 AM EDT
Connectivity temporarily appeared to return to normal, but we had not yet received an official restoration confirmation from Cogent and therefore continued monitoring.
UPDATE | September 3, 4:30 AM EDT
Several brief outages subsequently returned, with the longest lasting approximately 5 minutes. According to Cogent's latest update, the earlier recovery was temporary and field fiber repairs were still ongoing. Cable preparation was underway at one location, while conduit modifications were required at another before restoration could continue.
UPDATE | September 3, 5:00 AM EDT | Network Impact and Compensation
The incident did not cause the servers themselves to remain completely offline. Cogent's fiber failures affected specific network paths across multiple U.S. locations and some international traffic. Connectivity toward San Francisco and Hong Kong remained generally normal, while traffic to and from Cogent Los Angeles could experience unreachability, increased packet loss, or higher latency.
We are aware that some providers temporarily moved traffic to other carriers. However, LAX-BGP products are built around ISP-classified IP space supplied by Cogent. Announcing these IPs through another carrier would create new ASN history and change their network ownership and classification across third-party databases, affecting the ISP attributes committed to our customers and conflicting with our product terms. We therefore cannot use this approach.
Although this was a large-scale Cogent infrastructure incident rather than a NovixLink equipment failure, as the service provider directly responsible to our customers, we decided to provide compensation equal to 2 times the estimated downtime.
UPDATE | September 3, 8:00 AM EDT | Incident Largely Resolved
Cogent has confirmed that fiber repair at one location is complete, while end-to-end splicing remains in progress at the second location. Most affected connectivity has now been restored, and Cogent expects restored services to remain stable while the remaining work continues.
We therefore consider the incident largely resolved and will continue monitoring until Cogent formally confirms that all repairs are complete.
All affected LAX-BGP VPS services will receive a 2-day service extension. Based on observed availability, even twice the estimated downtime would amount to less than 2 days; however, we have decided to round this up to 2 full days as a goodwill credit. Compensation will be applied automatically after Cogent confirms the incident is fully resolved, and is expected to be completed on September 4 Beijing Time. No support ticket is required.
Unless Cogent experiences another material outage or its post-incident report contains important information not previously disclosed, this will be our final update for this incident.
We sincerely apologize again for the disruption and appreciate your patience and understanding.
Thank you for your business,
NovixLink
