METHODS TO IMPROVE THE FAULT-TOLERANCE OF INTERNET SERVICES

The article deals with the problem of ensuring the fault-tolerance of Internet services. It is becoming increasingly important due to the growth of the quantitative parameters of the functioning of information systems (users, servers, volumes of information in databases) and the level of complexity. Due to the heavy load on the servers on which the applications are deployed, there is a failure of hardware or software. The urgency of finding additional ways to ensure the fault tolerance, reliability and uninterrupted functioning of computer systems operating in real time, due to the increasing requirements for the security of running processes or objects, access to which is limited. Despite the use of existing methods for ensuring fault tolerance in cloud infrastructures, there is a problem of inconsistency between the actual availability of systems and the levels of "fault tolerance" (Fault Tolerance) and "High Availability" (High Availability) for critical and business critical web applications. Also, less attention is paid to methods of improving fault tolerance for highly loaded systems. Therefore, the search for new opportunities for scaling and load balancing when building cloud systems is of particular importance. To conduct a comparative analysis of load balancing algorithms, functional characteristics were proposed, according to which it is possible to make an informed choice of algorithms that most closely correspond to specific practical circumstances and the requirements of users and providers. The basic principles for implementing a resiliency strategy now include redundancy, diagnostics, reconfiguration, and recovery. As an additional indicator of failover control, it is proposed to apply the completeness of the booking of elements and components of the system, the completeness and reliability of the control, the probability of reserve recovery. The first two strategies assume that a system that correctly performs its functioning algorithm is safe. The safe behavior strategy for failures is used specifically for safe systems and consists in the transition of the system to a protective irreversible state in the event of a failure