Server designing method based on semi-synchronization, semi-synchronization and pipe filter mode

Abstract

The invention provides a server designing method based on semi-synchronization, semi-synchronization and a pipe filter mode. The method includes the steps of firstly, allowing a semi-synchronization layer to comprise a plurality of data processing modules; secondly, applying the pipe filer mode to the data processing modules of the semi-synchronization layer; thirdly, sequentially linking the data processing modules through a link table; fourthly, applying a binary exponential backoff policy to the application layer of a client. By the method, a whole server system has the advantages that system concurrency is increased greatly, and good system expansibility is achieved; the data processing modules can be dynamically added or deleted, and efficient and flexible data processing is achieved; server pressure can be effectively relieved by the binary exponential backoff policy, and server operation capability is increased at the same time.

Description

technical field [0001] The invention relates to a server design method, in particular to a server design method based on semi-synchronous, semi-asynchronous and pipe filter modes. Background technique [0002] With the continuous development of information technology, emerging computers, mobile phones and various media terminals are also emerging, which also makes it a challenge for these different terminals to communicate and cooperate with each other. As the core component of the distributed application, the network server's design will directly affect the performance of the entire distributed application. Although some relatively mature design patterns have been applied to the network server and have improved the performance of the server to a certain extent, such as the semi-synchronous and semi-asynchronous mode, etc., the currently used network server still has one or another. problem, the structure and disadvantages of these web servers are as follows: [0003] (1) ...

AI Technical Summary

Problems solved by technology

Therefore, this server model has the following disadvantages: a. Business logic and network logic are not separated, and the two will affect each other, which will also increase the probability of system errors; b. When a large number of network connection requests arrive at the same time, Will create the same number of threads as the number of network connection requests, which will take up a lot of system resources, and in the worst case may cause system paralysis; c, there is coupling between business logic and network logic, once the business logic changes, the system Large changes are required, which will increase the difficulty of system maintenance; d. If all threads are created in the same process, when any one of the threads fails, the entire application will crash; e. Frequent communication, which consumes a lot of system resources and leads to poor system performance

In this model, the number of threads in the thread pool is fixed, and a thread can provide services for different network connections at different times. Excessive load, but its disadvantage is: when the number of concurrent network connection requests is greater than the number of threads in the thread pool, it will not be able to process all network connection requests at the same time, which will cause some network connection requests to be blocked for a certain period of time

But this kind of network server also has the following disadvantages: a. When the business logic is more complicated (such as dealing with more complicated network application protocols), the thread pool strategy of the semi-synchronous layer is not flexible enough; b. The concurrency level is relatively single; c. Cannot dynamically increase and decrease business processing logic modules

Images