- After a reported drone strike on a VK data center, users and monitors recorded problems. VK says its services stayed available, and the cause of the outage is not established.
- The Record reports other data centers in Russia and Ukraine were also hit recently, so physical location is part of a provider's risk.
- Leaders should ask where their primary and backup systems physically sit, and whether failover has been tested across sites.
The cloud is a building with an address
People talk about online services as if they float. They do not. Photos, messages and songs sit on machines in a real place. If that place is attacked, the key question is how much of the service lived there.
That question arose in Russia over the weekend. The Record reports that thousands of users struggled to reach VKontakte, Russia's largest social network, and its music service. The trouble came after a Ukrainian drone strike on a VK data center in the Leningrad region, near St. Petersburg.
What is known, and what is not
VK confirmed that its facility was targeted in an overnight attack on October 18. It said its services were not affected. It did not say how bad the damage was or how many servers were involved.
Outside monitors saw something else. Downdetector logged about 6,600 problem reports on Sunday, around 600 of them in a single hour. The Russian service Sboy counted roughly 2,100 complaints. NetBlocks said VK-related connectivity fell to 56% of normal.
The Record says it was not clear whether the strike caused the disruption. That caveat matters. The two events share a night. The reporting does not show that one caused the other.
By Monday, complaints had fallen, according to the Russian monitor Detector404. Users still reported trouble uploading photos and videos, streaming music, logging in and sending messages.
Fire Point says this site mattered
Fire Point, the Ukrainian drone maker, said its FP-1 drones hit several targets in the region. One was the VK data center. The company described it as a key VKontakte site for storing and processing user data, such as photos and videos.
To show the size of the site, Fire Point cited earlier published figures: 640 racks holding over 20,000 servers. A rack is a standard cabinet for servers. These are Fire Point's figures, not numbers from VK or The Record.
Claims of heavy damage come from Fire Point, Ukrainian media and open-source analysts. The media and analysts described a partly collapsed roof. VK has not said how badly the building was hurt. The Record could not check those accounts itself.
Pavel Durov also spoke on Sunday. He is a VKontakte co-founder who left Russia and started Telegram. He said he built the site around fifteen years ago, and that VK took it over a few years later.
It is unknown how much of VK's service depended on this facility. VK denies that any of its services were hurt.
Data centers are now named targets
The Record places the VK strike in a wider run of attacks. Earlier in October, Ukrainian forces struck Yandex data centers three times. After the third reported attack, users across Russia had trouble reaching Yandex and many other websites. A Sberbank data center was also reportedly targeted.
Russia has hit Ukrainian sites as well. Its September attacks damaged at least six Ukrainian data centers. They included facilities run by MiroHost, Parkovyi, UTELS and Pavutyna. On Monday, Datagroup said a Russian ballistic missile destroyed equipment at one of its Kyiv data centers. It said the site was beyond repair.
Moscow has claimed that Ukrainian data centers supported the armed forces.
These reports cover one conflict. They do not show a global trend. They do show that, in this war, the buildings behind digital services are treated as targets.
How a service survives losing a building
Providers protect against a lost site by keeping a second one. Data is copied from the main site to the other site as it changes. If the main site fails, traffic is rerouted to the copy, and users may notice little.
That only works if the second site holds a full, current copy and can carry the load. If one building holds the only copy of storage or the login system, rerouting has nothing to point to. The service can be called redundant and still depend on one address.
The lesson: resilience lives at the address
Contracts promise uptime. Where the machines sit decides whether that promise holds. Whether this applies to VK, the reports do not say. For any provider, what matters is where each function physically runs.
The gap between VK's statement and the monitors teaches a second lesson. A company's own account of its availability is one input. Independent measurement is another. The two can differ.
Questions to put to your team
Ask where your primary systems and your backups physically sit. Ask whether any single site holds both. Ask which of your providers rely on one facility for storage, logins or payments.
Ask when failover to a second site was last tested. Ask what customers saw during that test. Ask whether you monitor your own services from outside your network. That way you learn of an outage from data, not from complaints.
Most organisations will never face a drone. Fire, flood, power failure and contractor error also destroy buildings. The VK episode is a reminder that every service has an address. Leaders should know where it is.
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: The Record.
