An Unusual Cloud: Researchers Propose a New Way to Make Data Storage More Reliable
Researchers at South Ural State University have developed a system that stores files across thousands of computers worldwide, making data storage 30 times more reliable than traditional cloud technologies.

In November 2026, researchers at South Ural State University (SUSU) unveiled a decentralized data storage system. Its defining feature is the elimination of a single dedicated server in favor of distributing encrypted file fragments across multiple personal computers. The technology is being developed as a foundation for fault-tolerant storage of documents and large datasets.
According to digital simulation results, this approach reduces the likelihood of data loss by more than 30 times compared with traditional cloud-based methods. One qualification is essential: the comparison is with baseline distribution algorithms, not a claim that the system is guaranteed to outperform every modern commercial data center. That distinction matters when assessing the technology objectively.

How the “Trust Score” and Integrity Checks Work
When a file is uploaded, it is split into fragments, encrypted and distributed across the network, with backup copies created along the way. The innovation lies in the algorithm used to assess the reliability of individual nodes. Each device receives a “trust score” based on its online availability, stability, response time and CPU load. The system does not treat a home computer that is switched off every night as equivalent to a reliable workstation. It also checks data integrity by calculating hash values for randomly selected fragments.
A hash value acts as a digital fingerprint: if a fragment has been altered, the check detects the discrepancy. Simulations showed that the network could redistribute data even when 30 percent of its nodes went offline simultaneously. In a real-world deployment, however, recovery time would depend on available network bandwidth.

The Research Background and Its Place in Russia’s IT Ecosystem
The SUSU project builds on a broader research trend. The university has been studying similar redundancy algorithms since 2022. In 2024, Delis, the company behind the Datananny file-storage system, announced that it was using spare capacity on office computers. In 2025 and 2026, the market also saw the emergence of scalable object storage systems, such as OpenYard’s Vortex Object, and geographically distributed solutions based on IPFS developed by HSE University.
Even so, SUSU’s new dynamic replication algorithm sets the project apart from comparable solutions. For Russia’s IT industry, it represents an important step toward developing sovereign systems software.
Legal and Technical Challenges of Decentralization
Developers and users of such systems must comply with the requirements of Federal Law No. 152-FZ, “On Personal Data.” Operations involving the personal information of Russian citizens must use databases whose servers are located in Russia. The Ministry of Digital Development has separately clarified how these requirements apply to stored document copies: the idea of “thousands of computers around the world” cannot be applied to Russian medical or government archives without a rigorous legal assessment. Other critical requirements include centralized management of encryption keys, guarantees that all copies can be permanently deleted, and protection against malicious control of multiple nodes. The “trust score” developed by the Chelyabinsk researchers measures only the technical reliability of computers; it does not establish whether their owners can be trusted. That problem requires a separate solution.

From Simulation to Real-World Pilots
At this stage, the technology has been patented as a computer program, but there is no publicly available evidence of industrial deployment or an independent audit. The most realistic scenario for 2027–2029 is a series of trials in controlled corporate or university networks. These pilots would make it possible to assess actual file read and write speeds, network overhead and total cost of ownership.
A thorough evaluation would also require specialists to obtain the developers’ simulation parameters, data on the number of nodes that were actually operating, and replication factors. Those details are needed to determine whether the performance gains come from the new algorithm or simply from increasing the number of copies. For now, decentralized storage is more likely to complement than replace conventional servers, enabling hybrid architectures designed to withstand failures.









































