Data gateway method and system based on distributed database

Abstract

The invention discloses a data gateway method and system based on a distributed database, and belongs to the field of distributed database systems. The method comprises the following specific steps: S1, establishing connection between a client and all database nodes of a back-end distributed database; s2, the data gateway carries out database protocol analysis on the SQL request data sent by the client; s3, combining the analyzed SQL request data with the working state of each node of the back-end distributed database to carry out load balancing calculation, and splitting the SQL request data; s4, distributing the SQL request data to corresponding database nodes for processing according to a calculation result of load balancing to obtain response data; s5, aggregating and sorting the response data according to the original distribution sequence, and returning the response data to the client; and when a fault occurs in a database node, the fault node is ignored by load balancing calculation, and request data is distributed to the normally working node, so that the client does not perceive the fault of the database.

Description

technical field [0001] The invention discloses a distributed database-based data gateway method and system, and relates to the technical field of distributed database systems. Background technique [0002] In a general peer-to-peer distributed database with Shared Nothing architecture, each node can work as an independent working node, and the client can connect to any node to access the database. When a node fails and cannot work, the client can Switch to another node to access the database, but the switch needs to be processed by the client. The database is not responsible for disaster recovery and fault migration functions, and the distributed database does not provide a unified access entry, so the client is required to achieve high availability. Increased user difficulty. If you use load balancing software, although you can provide a unified access portal, when a database node fails and cannot work, it will switch to other database nodes to work. However, the switchin...

AI Technical Summary

Problems solved by technology

However, the switch will require the client to re-establish the connection, and the data cannot be directly loaded to other database nodes. This requires the client to reconnect, which increases the processing difficulty of the client.

And the traditional load balancing software is based on the connection layer for load balancing, the load is the database connection, when the database connection is created according to the load balancing calculation result, the channel through which the data flows has been fixed, that is to say, the client All request data on the client side will be transmitted through this existing connection, and the database node receiving the request data is responsible for parsing the request data and generating scheduled tasks, which does not fully utilize the distributed database capabilities, because all the connections in a client The requested data is finally calculated on a database node. When the pressure is too high, the processing of this database node will be slow

In addition, the load balancing algorithm of traditional load balancing software is a relatively fixed algorithm and will not be adjusted according to the actual load pressure on the backend. When the pressure on a node is too high, the load balancing software cannot perceive it, or because of connection-based Load, there is no way to load data to other nodes for processing

Images