Cap theorem

Uniformity C Indicates that after the update operation is successful , All nodes have the same data at the same time . Strong consistency inevitably leads to inefficiency , For example , If the money from China Merchants Bank is transferred to Shanghai Pudong Development Bank , Both parties must make sure that the deduction received is successful before it is returned to the client , After the transfer, the interface will make us wait 5S, this 5S Time is when both systems are confirming .

Usability A When accessing data on behalf of a user , Whether the system can return the expected results in normal response time , For example, after the order system orders successfully , To pass the rpc Call the inventory system to deduct inventory , At this time, the operation of deducting inventory is more complex , Time consuming comparison field , In order to give users a better experience , We will asynchronously return a successful ack, Then let the user check , Ensure a good user experience .

Partition tolerance P Represents when a distributed system encounters a node or network partition failure , Still be able to provide external services to meet the consistency or availability . For example, the commonly used microservice multi cluster deployment is partition fault tolerance , That is, after a node goes down , Other nodes are available .

 AP It shows that after the order is created, the processing result is returned directly without waiting for inventory reduction

CP It shows that after the order is created, the result is returned only after the inventory is reduced

AC The performance is that the data system is no longer split , Complete the operation in a transaction of a database , That is monomer application