Perm Polytechnic scientists double website resilience to high loads with new approach
Researchers at Perm Polytechnic have developed a method for designing web services that doubles their ability to withstand heavy traffic. During trials, the new scheme cut RAM consumption and balanced CPU load, allowing sites to serve more visitors without failures, the university's press office said.
Digital platforms, including online stores, government services and mobile apps, process vast amounts of data daily, from financial transactions to doctor appointments. Any such service relies on a database that stores texts, images and personal user information. The site accesses it with every action: displaying a page requires retrieving data, while saving changes means writing it back.
The core issue is a limit on the number of simultaneous connections to the storage. As audience grows, the frequency of requests rises, and the database fails to handle them, causing delays, errors and complete unavailability. Such downtime translates into direct losses for businesses: customers cannot place orders, and companies lose revenue. For large enterprises, even a few minutes of outage can mean losses of hundreds of thousands of rubles. Failures also prevent citizens from handling everyday tasks like booking appointments, obtaining documents, paying taxes or calling emergency services.
Existing methods of reducing load typically distribute it across multiple servers or database replicas, but this requires extra resources and is not always effective. Perm Polytechnic scientists proposed a different approach — optimizing the service architecture itself. They developed an algorithm that redistributes the request flow across system components, preventing individual nodes from becoming overloaded. As a result, response time drops, and stability holds even with a sharp spike in user numbers.
The development targets high-load systems where uninterrupted operation is critical. The proposed approach, according to its authors, doubles resilience to peak loads without significant infrastructure investment, making it attractive for companies dealing with large customer flows.
