Methods of Increasing the Efficiency of Data Consistency in Information Systems

The article is devoted to special methods for distributed databases that allow to accelerate data reconciliation in information systems, such as IoT, heterogeneous multi-computer systems, analytical administrative management systems, financial systems, scientific management systems, etc. A method for ensuring data consistency using a transaction clock is proposed and the results of experimental research for the developed prototype of a financial system are demonstrated. The transaction clock receives transactions from client applications and stores them in appropriate queues. The queues are processed based on the transaction priority. The highest priority queue is processed before the lowest priority queue. This allows you to determine which important data (such as financial transactions) should be processed first. The article justifies the replacement of the Merkle tree with a hashing algorithm and the use of the Bloom spectral filter to improve the Active Anti-Entropy method to accelerate eventual consistency. For its effective use, the filter generation algorithm is modified, which allowed to increase the speed of its generation and maintain a sufficient level of collision resistance.