A Distributed and High Concurrency Message Matching Method

Abstract

The invention discloses a distributed high-concurrency message matching method, and specifically relates to the technical field of high-concurrency servers. The invention adopts a linear message queue arrangement method to realize data queues when data cannot be processed immediately. It is processed sequentially to ensure that each data can be processed, and with the leaky bucket algorithm, the cache of the lower layer of the processor can be adaptively allocated, and the parallel processing margin of the backup processor is adopted. Synchronous processing in the process of processing, forming a double-bucket strategy, with the cooperation of multiple backup servers for the database, can improve the data reading speed of the two processors without duplication of internal cache data in multiple intermediate layers, using multiple identical The method of synchronously reading different data from the data source guarantees the response speed of high access volume, and plays a decisive role in the stability of the database to adapt to the high reading state of high concurrent message matching.

Description

technical field [0001] The present invention relates to the technical field of high concurrency server, more specifically, the present invention relates to a distributed high concurrency message matching method. Background technique [0002] At the same time or in a very short period of time, a large number of requests arrive at the server, and each request requires the server to consume resources for processing and give corresponding feedback, such as the number of processes opened at the same time, the number of threads that can run at the same time, and the network connection The number, CPU operation, I / O, and memory are all limited, so the server can process requests at the same time is also limited. The essence of high concurrency is the limitation of resources. For example, if the number of online users in the system is 10W, it does not mean that the number of concurrent users in the system is 10W. There may be 10W users who view static articles on the homepage at the...

AI Technical Summary

Problems solved by technology

The essence of high concurrency is the limitation of resources. For example, if the number of online users in the system is 10W, it does not mean that the number of concurrent users in the system is 10W. There may be 10W users who view static articles on the homepage at the same time and do not send requests to the server. The high concurrency is based on the system The number of real users and sending requests require the server to consume resources for processing requests, and the server can only open 100 threads. Exactly one thread takes 1 second to process a request, so the server can only process 100 requests in 1 second

[0003] Under the access of large data, the database reads a large amount of data, which increases the pressure of data storage simultaneously. The security of data cannot be effectively guaranteed, and redundant requests cannot be processed. As for the high concurrency of data, its Data processing has been unable to achieve real-time processing, and at the same time, under the continuous backlog of a large amount of data, its database is under continuous high access volume, which puts a great pressure on the database server, and the response speed will also decrease, causing the processor to target The efficiency of high-concurrency message matching is further restricted, so it needs to be optimized