Database hierarchical storage optimization method capable of achieving flexible configuration

Abstract

The invention relates to a database hierarchical storage optimization method capable of achieving flexible configuration. The method includes the following steps that an SQL function is provided and the storage management mode of a user is provided, wherein the storage management mode is that a table, an index and a WAL log are flexibly configured according to needs; the storage management mode that the user utilizes the provided SQL function for customizing an own table, an index and a WAL log according to the active degree of application data is provided. The database hierarchical storage optimization method has the advantages that the hierarchical storage idea is adopted for optimizing traditional database storage management, a modified database not only supports NVM, SSD and other novel storage media, but also allows that the user stores the data according to an SQL instruction as needed in a hierarchical mode, in this way, the IO neck problem of a traditional database is effectively solved, and the performance of a whole system is remarkably improved.

Description

technical field [0001] The present invention relates to the technical field of database storage, relates to a database storage optimization method, in particular to a flexible configurable hierarchical database storage optimization method. Background technique [0002] Generally, the use of storage media mainly includes two categories: cache data: cache data for CPU high-speed calculation, such as DRAM; store data: permanently save application data, such as HDD; current mainstream storage media mainly include DRAM, NVM, SSD and HDD, Their characteristics are compared as shown in Table 1 below: [0003] Table 1 Comparison table of characteristics of current mainstream storage media [0004] [0005] Both DRAM and NVM in Table 1 are memory devices, which are expensive, and the access unit is bytes. They are often used to cache data during database operation. The main differences between the two include: DRAM and NVM have fast access speeds, and NVM has Non-volatile, but...

AI Technical Summary

Problems solved by technology

[0009] In terms of memory, traditional databases mainly use DRAM memory. Its main problems include: small capacity, unable to cache a large amount of calculation data, thereby limiting the processing power of the CPU; volatile, once the power is turned off, the data cached on DRAM may be lost. , so in order to ensure data reliability, the updated data should be stored in a permanent medium such as a hard disk in a timely manner, which can easily lead to CPU computing being blocked; in other words, the main problems of the current memory are small capacity and volatility

[0010] On the hard disk, traditional databases mainly use HDD hard disks. Its main problem is: although the storage capacity is very large, its IO access speed, especially the random access speed is very slow

[0011] To sum up, in the traditional DRAM and HDD-based method, users cannot design their own data storage solutions, and data storage is automatically realized within the system, that is, permanent data is stored on HDD hard disks, and cached data is stored on DRAM memory

The main disadvantages of this method include: due to the insufficient capacity of DRAM and HDD, it restricts the full use of CPU processing capacity

Images