Analysis of the Architecture of Distributed Systems for the Reduction of Loading High-Load Networks

In high-capacity networks, there is always a problem of delaying the receipt of packets between a client and a server. The load distribution should be made automatically based on the analysis of the distributed system state, since in the processing of Large Data it is necessary to analyze flows in a distributed, open dynamic system with a variable structure in real time. A distributed system for the task of reducing the load in high-capacity networks has been developed. An architectural scheme of “entering the remainder” is applied by introducing the new essence of the “last message”. This allows us to write the following message in the field of correspondence in the field. Therefore, we will be able to receive the latest message of any correspondence, but now, after each message arrives, it will be necessary to record it in two places. The cascade time synchronization scheme is proposed. The accuracy of time is important in distributed systems and allows you to synchronize the process. To do this, the Marzullo algorithm was used. This made it possible to establish a logarithmic relationship between the efficiency indicator and the number of machines. In this regard, it is important not to use too many computers with an algorithm that cannot provide efficient computer management. Improved messaging scheme. This allows you to define the entities used in this approach and to find references to each other. Query distribution managers send requests not only to each machine in sequence, but in real time recognize the one that is least downloaded and select it to handle the most demanding queries. This allows you to polynomically reduce the computation time.