Introduction

For many spheres of activity, such as banking, telecommunication the use of the distributed databases is more organic and effective method of organization of data storage and processing, than the use of the centralized systems data storage, it is conditioned business structure of organizations operating in these spheres. 

One of the most essential instruments of organization distributed DB is data replication. Replication of information is a mechanism of synchronization of content of a few copies of database. Replication can be synchronous or asynchronous. Asynchronous replication is used in the most real systems, when information after a change on one of sites, propagate on other sites during some time, but not right after a change. Replication also can be complete or partial. At complete replication — all of information are placed in all of databases. At partial — part of information is placed in part of databases, and to other, not replicated part of information, access is made

Objectives

The purpose of work is research of existent models of the perfomance of replicated and distributed databases. To one of important tasks, which it is planned to realize during implementation of work, there is experimental verification of rightness of model result information by the imitation of work of the high-loaded replicated database.

Related Work

To date the issues of the day is absence of universal algorithms and solutions for planning architectures of the distributed and replicated databases. To date practically all of architectures are tied to the concrete subject domains, that in particular is related to that many parameters of architecture can be acceptable in one area and quite unacceptable in other areas. Similarly a lot of limitations is laid on by the algorithms of conflicts resolution which are also tied to the concrete subject domains

Summary

Concluding, we believe that continuous effort in the development of advanced performance models for replicated
databases is needed so that they can keep pace with the evolution of distributed information systems. As one
example, the vision of «database access anywhere anytime» has several implications that influence the systems’
performance. Data distribution and replication continues to increase, wireless communication links suffer from
lower reliability and bandwidth, large database serves are replaced by clusters of workstations, etc. Such aspects
must be taken into account by future models and should be integrated with existing, proven modeling concepts.

Bibliography

1. Matthias Nicola, Matthias Jarke. Performance Modeling of Distributed and Replicated Databases.

2. Jim Gray, Pat Helland, Patrick O’Neil, Dennis Shasha. The Dangers of Replication and a Solution.

3. A.A. Volkov. TPC Tests, DBE # 2 p. 70-78.