Role of message queuing

The role of message queuing can be summarized as follows ： decoupling 、 asynchronous 、 Peak shaving . We use the current popular e-commerce system to illustrate the use of these three functions of message queue .

decoupling

If the upstream system calls the downstream system synchronously , It will greatly reduce the throughput and concurrency of the system , And the system coupling is too high . The function of decoupling is to reduce the system and the system , Or the coupling degree between modules in the system ,, Improve system throughput and concurrency .

The normal order placing process of e-commerce system is actually only to submit orders for users 、 Relationship Logistics , But for its background service , These processes will be handed over to “ Order service ” To complete , This set of actions may be related to operations involving inventory 、 Operation of marketing related functions 、 User points and other operations , And recommended or other operations . Before using message queuing , These related processes and operations are almost synchronized , Only when one step is completed can you continue to the next step , In this case, the coupling between various services or modules is very high , If there is a problem with the service of any module , Then the order placing process will not be completed . When using message queues , This problem can be completely solved . After the order service receives the user's request to submit an order , The task will be handed over to the message queue for processing , That is, the operation of commodity inventory 、 Operation of user points 、 The operation of marketing function and recommendation will be called by the message queue , For users , As long as the order is successful , It means that the process has been completed , Subsequent related operations will be performed without the user's knowledge .

Before accessing the message queue , The order service interacts with each module as shown in the figure above . After accessing the message queue , The order service will hand over the task to the message queue , The message queue is responsible for the interaction with each module . As shown in the figure below ：

asynchronous

After order payment , Back office service will deduct inventory 、 Add points 、 Sending messages and so on , And the processing flow is chain , A long link , The response time becomes longer . Import message queue , Basically, each node can handle it asynchronously , This reduces response time .

Peak shaving

Message queues can also be used to cut the peak , For example, seckill system , The usual traffic may be very low , But when it comes to second kill , The second kill moment will frantically request our server ,Redis,MySQL They have different bearing capacity , The underlying system may not be able to handle all the traffic directly . here , After introducing the message queue , We can throw a lot of requests into the message queue , Then take it easy , In this way, the goal of processing a large number of requests is achieved .

 

This article references from ： 20000 words 、 Thirty figures 、 Twenty three questions , Get it done RocketMQ! - Nuggets