A high-performance access method for relational databases in the financial field

Abstract

The invention provides a high-performance access method for a relational database in the financial field. The method comprises the following steps: S1) writing in data; S101) calculating the length of field through an application program, constructing a DML statement, taking internal storage as a cache mirror image, simulating a data table in the internal storage, filling into a data mirror image and sending the constructed DML statement and the data mirror image to a database management system through the application program; S102) analyzing the DML statement constructed by the application program by the database management system, controlling safety and integrity, controlling concurrence and transaction and sending a bottom data set to a disc of the database by the database management system; and S103) calculating address and offset, and then once writing the bottom data set into the disc of the database; and S2) reading the data. The invention has the beneficial effect that the access efficiency of the database is effectively increased.

Description

technical field [0001] The invention relates to a high-performance access method for a database, in particular to a high-performance access method for a relational database in the financial field. Background technique [0002] In the financial field, there are strict reliability and availability requirements for data storage, and the data in the financial field is extremely sensitive, with high requirements for access control, and in terms of data security and access efficiency in terms of technical system operation and maintenance , usually put the application program and the database in the same local area network. Mature relational data products, such as SQL Server and Oracle, are usually used in the current market, while the traditional access to data in relational databases is to construct DML-type SQL statements, and call packaged database API interfaces such as JDBC, ODBC, etc. , sent to the DBMS for parsing and execution, and the execution result is obtained. [00...

AI Technical Summary

Problems solved by technology

[0003] like figure 2 As shown, the traditional database access method is relatively flexible, safe and reliable, but even if the application program and the database are in the same local area network, the efficiency of database access is still low, because the main work is done in the DBMS, the bottleneck is also the DBMS, and the local area network The high-quality network bandwidth (between the application program and the DBMS) is often idle, except that it may occupy part of the uplink network link when the query returns a large result set, because the message packet composed of the DML type SQL statement is small , there is hardly any load on the downlink network link

According to repeated tests, even if the application, DBMS, and database are located on the same operating system platform, the maximum TPS of the application inserting data records into the tables in the database in the form of long connections is difficult to exceed 1500, and the TPS of modifying table data records is even lower. During the period, if the query load is added in combination with specific business scenarios, the efficiency of the DBMS will continue to decline.

Considering that quantitative / programmed trading methods are widely used in the current securities trading field, the performance of traditional database access methods can no longer meet the requirements, which limits the further development and expansion of the securities business. Therefore, it is urgent to find a solution that can effectively improve the efficiency of database access. method

Images