Yandex reported that attacks on several data centers between October 8 and 11 damaged infrastructure in the Ryazan, Kaluga, and Vladimir regions. The incidents disrupted access to some Yandex Cloud services and Yandex Disk. During recovery, engineers restored access to 97.16% of affected customer data. The remaining 2.84% could not be recovered because severe physical damage affected the storage equipment and no usable backup copies existed.
How much customer data did Yandex recover?
Yandex said it restored access to 97.16% of the affected data. It could not recover the remaining 2.84%, which was tied to storage equipment damaged beyond repair and lacked available backup copies. The lost data included workloads whose storage had sustained severe physical damage; logical recovery methods could not restore data where the underlying equipment and data integrity had been compromised.
Recovery teams first repaired access infrastructure, then checked storage systems for damage and data consistency. The work involved identifying affected equipment, restoring services where possible, and verifying recovered data. This sequence helped engineers bring systems back online while distinguishing recoverable information from data that had been permanently lost.
Which Yandex services and regions were affected?
The attacks damaged Yandex facilities in the Sasovo area of ββthe Ryazan region, as well as data centers in the Kaluga and Vladimir regions. Yandex Cloud and Yandex Disk users experienced access problems, delays, and, in some cases, short service interruptions. The company described a multi-day recovery effort involving repairs, infrastructure reconfiguration, and data checks.
Physical damage can affect cloud customers even when they do not manage the underlying servers. A cloud provider operates the hardware, but customers still need to consider how their data is backed up, replicated, and restored if a facility becomes unavailable.
Why did some customers avoid data loss?
According to Yandex, customers who used its backup service did not lose data in this incident. Backups can create recovery points through snapshots and incremental copies, while geographic replication can keep data in separate locations. These measures give customers options if a primary workload or data center becomes unavailable.
A snapshot captures a system’s state at a particular time. Incremental backups record changes since an earlier backup, reducing the amount of data copied each time. Geographic replication stores copies in another location, which can help maintain availability during a regional outage. The exact protection depends on the customer’s configuration, retention period, and recovery procedures.
What backup setting did Yandex change?
Yandex made backup protection enabled by default for newly created virtual machines, according to its announcement. Users can still adjust the setting. Existing customers should check their own VM configurations rather than assume that backups are active, since settings may differ across older workloads and individual projects.
Before relying on a default, confirm which resources it covers, how often it creates recovery points, how long it retains them, and where it stores copies. A backup that exists but cannot be restored within the required time may not meet a business’s recovery needs.
How should cloud customers prepare for a data center outage?
Customers can reduce the impact of equipment damage or a regional interruption by combining independent backups, geographic separation, and tested recovery procedures. Start with the systems and data that would cause the greatest operational or financial impact if unavailable.
- Classify critical workloads: Identify databases, business applications, and customer files that need the strongest protection.
- Set backup schedules and retention: Choose snapshot frequency and retention periods based on how much recent data the business can afford to lose.
- Keep copies in separate locations: Use geographic replication or independent backups so a single facility outage does not remove every copy.
- Test restoration: Regularly restore a workload from backup and confirm that the data and application function correctly.
- Define recovery targets: Set a recovery time objective (RTO), the maximum acceptable time to restore a service, and a recovery point objective (RPO), the maximum acceptable amount of data loss measured in time.
- Monitor backup jobs: Alert teams when copies fail, fall behind schedule, or become unavailable.
How can a business test its recovery plan?
A practical recovery exercise checks more than whether a backup file exists. It should confirm that the team can access the copy, restore the data, restart the application, and verify that users can complete essential tasks.
- Step 1: Select a critical service and document its dependencies, such as databases, network settings, and access credentials.
- Step 2: Restore the service into a separate test environment using a recent backup or replica.
- Step 3: Check data consistency and run representative application tasks.
- Step 2: 4: Record the time required and compare it with the business’s RTO and RPO targets.
- Step 5: Update procedures to address missing credentials, unclear responsibilities, or restore delays.
For example, a business that takes daily snapshots may still lose close to a day of recent changes if it restores only the latest snapshot. More frequent recovery points can reduce that gap, while a separate replica may help the service resume sooner. The right setup depends on the workload and the consequences of interruption.
What does the incident show about cloud resilience?
Cloud infrastructure does not eliminate physical risk. It shifts responsibility for equipment and facility operations to the provider, while customers remain responsible for choosing suitable backup, retention, and recovery arrangements. Geographic distribution can reduce dependence on one site, but customers should verify that copies are genuinely independent and that their applications can operate from another location.
Yandex said it planned to continue distributing computing resources across data centers in different Russian regions and to increase the number of sites. The company also described measures such as geographic redundancy and more frequent hardware health checks. These steps can improve resilience, but customers should still validate their own recovery plans and service requirements.
Key takeaways for Yandex Cloud users
Yandex reported recovering access to 97.16% of affected customer data, while 2.84% remained unrecoverable because of severe physical damage and unavailable backup copies. Customers using Yandex’s backup service were reported to have avoided data loss. To prepare for future outages, users should confirm that backups cover every critical workload, keep copies in separate locations, and test full restores on a regular schedule.
Be the first to comment