There are two main areas of research database systems. One of them &ndash an analysis of data models, which can display real world and on which can build interfaces for different types of users. Such conceptual models include hierarchical, network, relational model, as well as a number of models focused on semantics.
The second area involves the safe and efficient implementation of the DBMS. Computerized data are becoming a central resource for most organizations. This should be taken into account in each implementation, intended for production use, by ensuring data security in case of concurrent access, recovery and restructuring. One of the major criticisms of many early database belonged to a lack of efficiency in the processing of their proposed high-power operations, especially data access based on their content across requests. Query optimization is designed to solve this problem by integrating a large number of methods and strategies, ranging from the logical query transformations to optimize access paths and storage at the file systems.
The problem of query optimization requires the following tasks:
Converting to more efficient query nonprocedural representation (logic optimization).
Choice of alternative procedural query execution plans.
Estimating the cost of the query on the selected plan, etc.
For every problem there is more than one approach to solving them. For example, the problems associated with the logical query optimization, gave rise to the direction, called semantic optimization [1]. Too many researchers are problems valuations procedural query execution plans (and still question the reliability of estimates is not completely clear) [2].
Especially a lot of research in recent years devoted to query optimization, and choosing effective ways to perform relational operations in a distributed relational database management systems [3]. Here there are so many options and the physical organization of distributed databases (with support for copies of the relationships within the network nodes, with horizontal or vertical separation relationships in multiple sites, with support for snapshots of the database, etc.) and algorithms to perform relational operations in each organization.
Objectives of work:
Developing a model system of choosing the optimal global plan of a distributed query.
Researching and selection of the parameters of the system.
Developing an algorithm based on ant colonies to determine the optimal global query plan.
Optimizing database queries is the most important and interesting area of ??research and development in the entire database. The importance of this direction is determined by that from the developed component of query optimization depends critically on the overall performance of any SQL-oriented DBMS. This trend is most interesting, because the solution of optimization problems can use a variety of approaches and methods from various areas of computer science and mathematics: methods for optimizing the programs used in compilers, programming languages, mathematical logic, mathematical statistics, methods of artificial intelligence, pattern recognition, etc . etc.
Most modern optimizer uses exhaustive search method for forming and selecting the optimal global plan of a distributed query. For small distributed database is justified because the sheer number of options are not so great. Recently, however, are gaining popularity distributed systems with many nodes (eg, cloud technology) for which an exhaustive search of variants can significantly affect performance.
The urgency of this work is to use the algorithm does not perform an exhaustive search, but returns one of the best results.
Scientific novelty:
On the basis of the model of the graph of operations (transitions) was developed a unique graph model of the optimization process of the global query. In progress Master's thesis will investigate its options.
First used ant algorithm to solve the problem of query optimization.