PROBABILISTIC MODEL FOR ESTIMATION OF CAP-GUARANTEES FOR DISTRIBUTED DATASTORE

Abstract

The subject of the article’s research is the CAP-guarantees of distributed datastore. The goal is to evolve decision-making algorithm for the distributed datastore architecture design which will balance CAP-guarantees depending on business requirements. To achieve that the following problems were solved in the paper: the stochastic model to evaluate different components of CAP-characteristics and some metrics that will impact on these values were developed. To solve these problems the following methods were used: basics from graph theory and probability theory, general formulas of expected value and automaton models and software application for calculation of developed formulas. The capability to measure such metrics resulted in to forming some constitutes of decision-making algorithm. Conclusions: the developed components of decision-making algorithm were the purpose of this paper and it could be one of basic components on the design distributed datastores stage, so that architects who build new software design may also use the algorithm to achieve balanced guarantees of distributed system reliability at the earlier stage of business needs implementation.

Keywords

• CAP-guarantees
• distributed data warehouses
• decision making algorithm
• stochastic metrics
• average replica propagation time
• design of distributed data warehouses