Why the failures of Yandex data centers became an alarming signal for Uzbekistan's business

A person opens an app to call a taxi or pick up an order and gets an error message. For them, it's wasted time. For the driver, it's a missed trip. For the seller, it's a delayed sale. If there are thousands of such users, a technical problem turns into an economic one.

This is precisely why attacks on Yandex's data centers are significant far beyond Russia. They have shown how much everyday services depend on equipment that can be located in another country, and how important it is to prepare for its shutdown in advance.

A data center is a building with a large number of servers: computers that ensure the operation of websites and applications. When a company uses the 'cloud,' it rents such computers and related services from a provider. Behind the word 'cloud' are always real buildings, power supply, and communication lines.

What happened

Over several days, Yandex reported attacks on three sites. On the night of October 7-8, a data center in Sasovo was severely damaged, and its operation was halted. On October 9, an attack affected a site in Kaluga, where some equipment failed. On the night of October 11, a data center in Vladimir was attacked, and its operation was also stopped.

According to the company, employees were not injured. The timeline for restoring equipment at the time of the statement remained uncertain. However, confirmed information does not allow us to say that all of Yandex's data centers have been completely destroyed.Yandex's official statement.

Yandex Cloud reported the unavailability of two zones of its infrastructure and the platform's transition to emergency mode. A zone is a separate part of the cloud used, among other things, for distributing equipment among sites. The provider also warned about restrictions on creating new resources in other zones: after the incident, there might not be enough free capacity for everyone.Yandex Cloud's announcements.

It is important to distinguish between a service outage and data loss. If an application does not open, it does not necessarily mean that orders, documents, or money have disappeared. Such conclusions require separate confirmation.

How it affected Uzbekistan

On October 8, Uzum Market experienced a failure that affected the marketplace, warehouses, and pickup points. However, it cannot be definitively considered a direct consequence of the attack on Yandex.

In a comment to Spot, Uzum's press service explained the disruptions as planned system updates. The company stated that it uses its own equipment and cloud servers within Uzbekistan. Uzum Bank announced that it is operating normally and the marketplace failure did not affect its services.Spot's material with company comments.

On October 11, disruptions were also observed in some Yandex Go services in Uzbekistan. According to UzDaily, Yandex Uzbekistan representatives, at the time of publication, did not specify whether the local problems were related to the incident in Russia.UzDaily's report.

Therefore, it cannot be claimed that Uzum and the Uzbek taxi service entirely rely on Yandex Cloud and stopped for the same reason. But these events raise a common question for businesses: how to serve customers if the main computer system stops working?

Why a major provider doesn't guarantee uninterrupted operation

Renting equipment from a large technology company is convenient. Businesses don't need to build their own data center, maintain servers, or purchase new computers as demand grows.

However, provider reliability and customer readiness for an emergency are different things.

Imagine a retail chain that stores all its goods in one warehouse. The warehouse may be modern and well-protected. But if it's closed, the stores will be left without supplies. A second warehouse only helps if it already has the necessary goods and delivery is organized.

It's similar with digital services. Backup servers will help if they have the necessary data and are capable of serving customers. If all this still needs to be organized after an incident, recovery will take time.

Even multiple sites do not always provide complete protection. They may depend on the same management systems or face a common threat simultaneously. Yandex Cloud itself warns: after one zone fails, free resources in the others are consumed faster.Provider recommendations.

What could have reduced the impact

Without access to Uzum's and Yandex Go's internal systems, it's impossible to determine what measures they already had in place and what exactly would have prevented a specific failure. But for companies with similar risks, there are several practical solutions.

First is to have a prepared backup site.

A company can place a backup with another provider or use its own equipment. The main thing is to ensure it can operate when the primary system is down.

Not every business needs two equally powerful sites. Sometimes, a small backup system that supports core operations is sufficient: displaying processed orders, saving customer inquiries, and helping to complete ongoing services. If necessary, its capacity can be increased.

But the backup must be prepared in advance. A contract with another provider does not automatically mean the application can be launched quickly.AWS backup approaches in recommendations.

Second is to store separate copies of important data and verify them.

Copies of orders, documents, and other necessary information should survive the primary site's shutdown. It's not enough to receive a 'copy complete' notification. You need to periodically check if work can actually be restored from these copies.

This is like a spare key: its existence is only useful when it fits the lock and is available when needed.

For payments and orders, accuracy is especially important. After recovery, the system should not charge twice, re-process a purchase, or lose an already confirmed transaction.

Third is to plan for limited functionality.

During an incident, it's not always necessary to disable the entire application immediately. Sometimes, non-essential features can be temporarily removed while preserving core functions.

For a marketplace, access to an already placed order is more important than personalized recommendations. For a taxi service, the ability to correctly complete a current trip is more important than additional offers in the app.

However, there's a limit: if the system cannot reliably verify payment or product availability, accepting new transactions is risky. It's better to explain a temporary limitation than to make a promise to the customer that the company cannot fulfill.

Fourth is to implement updates cautiously.

The reason Uzum cited for the failure suggests that an outage can occur even without data center damage. An error in a new software version can also disrupt business operations.

Therefore, changes should first be tested on a small part of the system. If problems arise, the team should have the ability to halt the update and revert to a working version.

Routine customer and employee actions must be tested: placing an order, accepting a delivery, issuing a product. If the main page opens, it doesn't mean everything is working.

Fifth is to conduct drills before a real incident.

Management needs clear answers to two questions: how long will recovery take, and what are the latest operations that might not be saved?

For a company information website, a few hours of downtime might be acceptable. For a payment service, the requirements will be much stricter.

Answers must be confirmed by practice. During drills, the team tests the backup site, data recovery, and actual user actions. This reveals problems that are not apparent on a normal day: missing access, outdated copies, or insufficient capacity.

Sixth is to plan customer communication in advance.

During an outage, people need specifics: can they receive their order, is their payment saved, do they need to repeat the transaction, and where will the next update be available?

For sellers and partners, clear rules for rescheduling deliveries and handling delays are also important. Silence increases the load on support and prompts customers to take repetitive actions that can complicate recovery.

Why can't everything just be moved abroad?

The choice of a backup site depends on more than just distance. Businesses in Uzbekistan need to consider requirements for storing and transferring personal data, and regulated industries have additional rules. The admissibility of a specific solution must be verified considering the type of information being processed.Uzbekistan's Law "On Personal Data".

One option could be independent sites within the country. But even here, verification is important: do they use different power sources and communication channels, can they operate independently of each other?

The cost of disaster preparedness

Backup infrastructure requires money. But downtime also has a price: missed sales, compensation, delayed logistics, additional employee work, and loss of trust.

It's important not to consider the entire volume of canceled orders as a final loss – some purchases are rescheduled. Management needs to assess the real consequences of an outage and choose protection commensurate with their business.

You can start simply: determine which systems are essential for serving customers, check data backups, and conduct a trial recovery. Then, prepare a backup of core functions.

The events of October have shown that behind convenient applications lies vulnerable physical infrastructure. Customers cannot influence the design of data centers, but they have the right to expect that a company will prepare for their shutdown. For businesses, such readiness becomes part of the service quality – along with price, delivery speed, and app convenience.