Research and construction of stress testing DBMS models to increase speed and performance of distributed computing system

Modern information systems work with huge data flows, so there is a need to build a logically single database (DB), parts of which are located within the network nodes. Selecting a DBMS node of distributed system is a complex multiparameter task and is one of the important stages in the development of DDB. The choice of the DBMS is determined by many different factors such as the number of possible users of the system, performance and reliability. The aim is to identify the criteria for building the DBMS testing algorithms followed by its testing and use as a node in a distributed database increases the speed and performance of the entire system. In the framework of examining the use of DDBs in the site freely available databases such as MySQL and PostgreSQL. The most important criteria for comparison, from the point of view of the user, are performance and reliability. By performance we mean the value that the database spends on particular data processing operations. The main factor influencing the performance is response time. In turn, the response time can be divided into: the response time for inserting a single element; response time to remove one element; response time to read the entire table; response time to read the mask and *response time to update a single element; response time to insert elements from the 50 % N, where N - initial number of records in the database; response time to clean up the table. The concept of reliability of the system has a lot of meanings. The work under the reliability we mean the safety of information transmission. To carry out a comparative analysis a set of tests analyzing the DBMS performance of MySQL and PostgreSQL was developed. The developed complex of tests includes: database test algorithm response time for data entry; DBMS test algorithm response time when reading information; DBMS testing algorithm performance; DBMS testing algorithm for reliability. To conduct a comparative analysis of the above selected criteria the planned experiment and the regression model was carried out. The study dependencies, built on regression models, allows to say that PostrgreSQL is slower when inserting data into a table, but the database is much faster at reading data tables. Just PostrgreSQL DBMS runs faster with highly filled base and a large number of users. Thus, within the framework of the nodes DDB it is better to use PostgreSQL.