Three Sites In Four Days
Between the night of 7 to 8 October and the morning of 11 October, drone strikes hit three of Yandex's five Russian data centres, The Register reports, at Sasovo, Kaluga and Vladimir.
| Date | Site | Reported state |
|---|---|---|
| 7 to 8 October | Sasovo | Seriously damaged and shut down completely, with a fire; two of three AI supercomputers there out of service |
| 9 October | Kaluga | Part of the infrastructure offline, operations degraded |
| 11 October | Vladimir | Supports Yandex Pay; offline, reported as stopped completely |
| 12 October | Yandex Cloud | Limited mode, customers advised to use alternative sites |
The Register and BleepingComputer attribute the strikes to Ukraine. On the afternoon of 11 October heise noted that Ukraine's defence ministry had not commented, and Yandex reported no injuries.
What Yandex Told Its Cloud Customers
Yandex's own status page described the cloud platform's mode as an emergency and recommended using any other alternative disaster recovery plans.
It also called the remaining resource configuration unstable and warned that technical support channels could be unstable. The Register adds that some customers can reach the console and the API but not consistently, that support answers more slowly because of the volume of requests, and that Yandex recommends alternative sites for hosting and reserving resources.
BleepingComputer's reading of the status page lists the console, API, identity management, container registry, databases, S3 storage and Kubernetes as available with some performance or availability problems, compute, AI Studio and security services as partly available and Monium as offline.
NetBlocks measured about 80 percent of connections to Yandex servers working by late morning on 11 October and about 30 percent of connections to Yandex Cloud servers. Yandex says its engineers restored Alice AI, Mail, Maps, Search and Taxi first, while its Yandex Pay payments service stayed offline a day after the Vladimir strike. The head of its fintech division wrote that users' money is safe.
What Nobody Has Confirmed Yet
Whether any customer data was lost for good is unclear, and Yandex has not addressed it in its status updates.
One customer posting on Habr described Yandex Disk backups that lost their thumbnails and files that failed to download, with HTTP 302, 504 and 500 errors, starting on 8 October. That is a single report, not a measurement.
Yandex says it cannot provide a reliable service restoration timeline and that its teams are working around the clock.
Ukraine's Data Centres Were Hit In September Too
At the end of September Russian strikes damaged the data centres of Vodafone Ukraine, Datagroup, Cosmonova, Omega Telecom and DreamLine and hit the headquarters of Kyivstar, according to Reuters as cited by heise.
Reuters reported that Russia justified such targets by saying the data centres had links to the Ukrainian army. The point for a buyer in Europe or the UK is operational: physical sites that carry other companies' workloads have been struck in an active conflict, and a provider with few sites has few places to move to.
Questions To Put To Your Own Cloud Provider
Ask how many independent sites sit behind the availability promise in your contract, and how far apart they are. Yandex lost three of five sites in four days, which shows what a small site count means in the worst week.
Ask where your backups live. Backups kept with the same provider, or in the same country, can fail together with the primary copy. A copy at a second provider in a different jurisdiction is the one that survives.
Ask for the time of the last tested restore, measured in a real drill, and read the force majeure clause to see what the provider owes you when a site is physically destroyed.
Read next: 21 Months of Notice Did Not Save German Clinic Links | Three Windows Support Deadlines Land Within 30 Days
