Grigor'ev

The technique of application of the program Complex instruments of Analysis Tools Models access to databases ( CCAY ) distributed processing systems ( RSOD ), developed on of the Chair "Information processing and management " Bauman . NE Bauman with Amur  State University, is considered. It is intended for carrying out computational experiments to analyze the temporal parameters of the designed information systems (IS ), wasps nova which are distributed databases and applications. Development environment - Personal Oracle and Developer 2000.

We consider the purpose of the complex, which belongs to a class of expert systems and is designed for computing ( model ) experiments to analyze the performance indexes (time performance ) of distributed processing systems ( RSOD ) . A brief description of the subsystems of complex parameters and inputs. The software package was used in the design of information systems, and is also used in the educational process at the Department "Information processing and management," Bauman. NE Bauman. It is used by students when the course work in the discipline of "Structural Design of the information processing system".

Known architectures of parallel database systems are analyzed in this article. The authors also consider expressions for evaluating the mean time of connection of two tables by NLJ and HJ methods; relevant mathematical expectations for SE, SD, SN architectures of parallel database systems are also considered. The authors provide examples of query processing in the following resources: disk array, RAM, connecting bus and processor. If there is a bottleneck in the resource, processing in that resource can be simulated by using the M/M/1 queue system. Parameters of that queue system are evaluated. The example of calculation which allowed to make several non-trivial conclusions is considered in the article.

The article gives an expression for mathematical expectation of the execution time of data warehouse inquiry built on the basis of a database parallel system (DPS). The author determines parameters of this expression taking into account the presence  of “bottleneck”. The author provides an expression which makes it possible to estimate the mean execution time of inquiries to the database parallel system with CE cluster architecture. In this expression the number of SMP-units and quantity of AMP-processors in each unit are taken into account. A practical example is analyzed, and conclusions are formulated. In particular, it is noted that if “the bottleneck” (disk) of the system is not overloaded (its load is less than 1), then the mean execution time of data warehouse inquiry in the system with the SMP-units is commensurate with the time of processing this demand in the MPP-system with the same quantity of AMP-processors. This makes it possible to substantially save cash resources during the acquisition of hardware of the parallel database system.

Concepts of complex that belongs to class of expert systems and designed for conduction of numeric experiments (simulation) in order to analyze productivity (timing data) indexes of distributed systems of data processing (DSDP) were considered. New class of analysis model of DSDP’s timing performances was developed in this work with optimization mechanisms of queries to distributed database taken in the account. Application of the decision-making process, based on holistic selection of architecture type was also developed. If frame of expert system is connected with SQL operator, then frames unite into network that corresponds to the sequence of SQL operators’ execution. Different resources of computer network are used when SQL operators are executed that is frame network goes into resources of computer network. 

This article deals with the development of databases of ADSDP subsystem of software solution for analysis of productivity indexes in distributed systems of data processing, designed to analyze computer networks. Different problems were solved while developing schemes of databases for ADSDP; they included description of nodes, channels, DSDP topology, virtual communication channels and their characteristics; determination of node configuration and channel configuration based on the СMAT registries; distribution of transactions with description of SQL-queries and tables of DSDP databases over nodes; storage and maintenance of components of devices’ load ratio and execution time of transactions (these components are used for maintenance of ES explanation mechanism); maintenance of ADSDP data base integrity.

In the article the authors suggest expressions which make it possible to estimate the aggregate value of possessing database parallel systems (DPS). The authors developed an algorithm of DPS selection which allows for specific character of the comparison of DPS architectures and special feature of cost estimates. By means of sequential growth of the number of AMP-processors and SMP-units the algorithm orders the database parallel systems with SE, CE and SN architectures according to their cost increase. It is shown that in this sequence it is possible to isolate the subsequence in which DPS are ordered according to descending ordering of the mean time of responding to requests. As in the algorithm all DPS versions are consecutively analyzed, in the finite number of iterations optimal solution will be found

Mathematical apparatus of analysis of SELECT-query execution time in distributed database, considering the mechanisms of complex query decomposition into sub-queries and connection of their results, the parameters of logic scheme and random characteristics of database filling. Proposed approach could be used at early stages of design for estimation of time characteristics of execution of different complex queries on different hardware platforms. In this manner, estimation was in form of Laplase-Stieltjes transformation, that allowed to estimate not only average values but also dispersions and other issues.

This article deals with algorithm of synthesis of database scheme based on the given set of functional dependences. Bernshtein’s algorithm was described; it was shown that usage of (G* + J)+ = (G + J)+ condition caused calculation problems, because covering of FD F+ set could exponentially depend on the size of F. Algorithm, based on Ullman’s ideas, was proposed; it provided relatively simple implementation. Generated database scheme had the property of lossless connection, maintenance of functional dependences and each sub-scheme of this database was in the third normal form. Example of database scheme synthesis using the proposed algorithm was included in this article.