Intelligent control method for congestion of instantaneous high-concurrency server thread pool

Abstract

According to the intelligent congestion control method for the instant high-concurrency server thread pool, short response time can still be kept in a server system with large instant concurrency, and the intelligent congestion control method for the instant high-concurrency server thread pool can be realized by analyzing the operation mechanism and the implementation principle of the thread pool and the congestion control algorithm and clearing the feasibility of implementation of the congestion control algorithm in the thread pool. The method comprises the following steps: analyzing and comparing action parameters of a thread pool and a congestion control algorithm, determining conditions for selecting the congestion control algorithm, analyzing the congestion control algorithm, selecting an algorithm meeting the conditions, improving the algorithm in combination with characteristics of the selected congestion control algorithm meeting the application of the thread pool, and realizing the algorithm in the thread pool. According to the method, the specific method for applying the action parameters in the congestion adaptive control to the thread pool is provided, the method has good adaptive capacity to the server with large instantaneous concurrency, short response time can be guaranteed, and the method has high expandability, high practical value and wide application prospects.

Description

technical field [0001] The invention relates to an intelligent control method for thread pool congestion, in particular to an intelligent control method for instantaneous high-concurrency server thread pool congestion, and belongs to the technical field of high-concurrency thread pool congestion control. Background technique [0002] With the rapid development of network technology, various communication exchanges are becoming more and more frequent, and instantaneous high-concurrency task requests often arrive at the server at the same time. Excessive logic processing and I / O read and write operations have brought great challenges to the server's load processing, and it is urgent to improve the server's instantaneous high-concurrency processing performance. The server should have the characteristics of low response time and high throughput, and adopt multi-threaded programming. Multi-threaded programming adopts multi-threaded concurrent processing. Different threads perform...

AI Technical Summary

Problems solved by technology

[0009] First, there are some problems in the application of the thread pool technology in the prior art, among which the more prominent problems include: first, it cannot be well applied in the server with instantaneous high concurrency, the ability to adjust the number of threads according to the request is poor, and it cannot Respond to user requests in a timely manner, and even lead to request loss; second, the algorithmic statistics and prediction methods consume more resources, and at the same time cannot cope with emergencies; third, it cannot respond to concurrent instantaneous requests, and the priority proposed at the same time does not consider low priority Level-long waiting and other problems have limitations in use; fourth, it is obtained based on a specific research background, and its application scenarios have certain limitations; fifth, it is necessary to count user access frequency, so it is necessary to store historical data and make statistics, which consumes a lot of Multiple resources, and statistics also need to be implemented in a more complicated way;

[0010] Second, a wide variety of network applications pose a huge challenge to network services. The thread pool of the existing technology cannot respond in time to the system with a large amount of instantaneous concurrency, which reduces the response speed and increases the response time to requests, which cannot satisfy users. Requirements, poor adaptability to servers with large instantaneous concurrency, and response time cannot be guaranteed; it is not conducive to network services becoming efficient, low response time, high throughput, and saving service resources;

[0011] Third, the dynamic thread pool of the existing technology has certain limitations. When a large number of users access concurrently at the same time, the number of threads will not increase in time. At this time, the number of tasks will be in a waiting state, and too long waiting will cause The task is discarded, which eventually leads to the occurrence of class congestion. When the server does not respond in time and discards the task, it will call related functions to process the task, and feed back the congestion situation of the system to the server. It is necessary to propose a thread pool congestion for the system. Control method, congestion control adaptively adjusts the rate of the sending end according to the network condition, when the network condition is good, the congestion control increases the sending rate, and reaches the maximum capacity of the network channel as soon as possible; when the network is congested according to the feedback information, the congestion control controls the sending end through Process, adjust the sending rate, and reduce the occurrence of network congestion;

[0012] Fourth, although the thread pool dynamic algorithm based on the average number in the prior art is stable, the predicted thread pool size results are all inferred based on the previous stage, which has hysteresis. In the face of instantaneous concurrent applications It is inevitable that there will be problems in the environment; the adjustment of the segment-based thread pool dynamic adjustment algorithm is an improvement on the first algorithm, which is stable, but when making predictions, the last result is used as the adjustment base, resulting in a gentle increase , there is also the problem of untimely response to the situation of large instantaneous concurrency; the advantage of the intelligent algorithm based on the average processing time is that it can directly compare the average task processing time with the previous time according to the calculation, and fine-tune it if it does not meet the requirements within a certain range Meet the required requirements, but if faced with instantaneous concurrency tasks, due to the large time gap, its adjustment will take too much time and it is difficult to meet the requirements

Images