Summary
- Yandex reported partial loss of operational modules at its Kaluga data centre after a drone strike.
- The Kaluga incident is separate from the Sasovo outage already covered by DataCentral.
- Precise equipment damage, workload recovery and full customer consequences remain incompletely disclosed.
Yandex reported that a drone strike partly disabled its data centre in Russia’s Kaluga region on 9 October, after a separate strike on the company’s Sasovo facility had already disrupted operations. The second incident affected a separate physical location, extending the operator’s infrastructure disruption rather than simply prolonging the earlier outage.
Reuters reported the Kaluga strike on 9 October, citing the company and contemporaneous accounts. Some modules were put out of action, but the precise systems damaged and the number of workloads affected were not established publicly at the time. Descriptions of the broader Yandex platform’s service difficulties should not automatically be assigned to the Kaluga site alone.
A second facility changes the recovery problem
When a data centre loses part of its usable capacity, the recovery task depends on which components remain operational. A damaged power path can interrupt otherwise undamaged servers; a loss of network or storage equipment can affect services even when computing hardware still has electrical supply. Yandex has not released enough technical information to attribute the Kaluga outage to one such mechanism.
The separate Sasovo strike is significant because data centre resilience frequently depends on the ability to move services away from a failed location. Multiple damaged facilities make it harder to assume that an alternative site will always have spare power, storage, connectivity and compatible compute capacity. That is a general operational consideration, not confirmation of a particular failure in Yandex’s recovery architecture.
Customer-visible effects vary according to application design. Services with copies of data and independently provisioned computing resources can sometimes restore operation elsewhere. Systems that depend on tightly coupled storage, specialist accelerators or specific network routes can face more complicated recovery procedures. Yandex has not provided a verified service-by-service recovery schedule for Kaluga.
Damage assessments remain incomplete
Contemporaneous reporting identified the Kaluga campus as a large site that supports Yandex services, but a reported rack count or campus power rating should not be confused with capacity lost in the incident. The disclosed result was that part of the facility was put out of action, not that every system on the site had failed.
The attacks occurred amid escalating strikes on physical infrastructure in Russia and Ukraine. The technical story concerns the interruption of production computing facilities and the resulting demands on redundancy, service restoration and safe site access. It is not possible to infer from the limited public evidence which protective measures were installed or why particular equipment sustained damage.
Engineers also need to establish that damaged electrical and cooling equipment can safely be re-energised before bringing sensitive computing systems back into service. Even where applications are restored from another location, the original campus may require a different, longer process of inspection and repair.
Reuters identified Kaluga as the second Yandex data centre struck within two days. The company operates a distributed estate, but the number of locations alone does not establish whether affected applications had usable spare capacity in another region. Availability depends on where live replicas were maintained and whether the target application can restart without its original storage or processing hardware.
The next material disclosure would identify which modules can return to service, whether data and hardware are recoverable, and how Yandex is allocating remaining capacity. Without that information, the public record supports a partial Kaluga data centre shutdown and unresolved damage, but no definitive estimate of capacity permanently lost.
