An example analysis of the CAP theorem:

Let's talk about the conclusion first: the CAP theorem says that we can only satisfy two of consistency, availability and partition tolerance.Among them, partition fault tolerance must be satisfied, so we have to make a choice between consistency and availability.

(1) The banking system is a server cluster with multiple servers to improve performance

(2) At first, Xiaoye stored 50 million in the banking system. At this moment, a server was offline and online due to network turbulence.At this time, data synchronization is performed again.

(3) A question arises: in the process of synchronizing data, is the entire banking system allowed to provide external access?

If it is allowed, then we default to availability, but it is possible that the data is not synchronized and the consistency is not fully established.Therefore, it can only satisfy the partition fault tolerance and availability.

If it is not allowed, then we will default to consistency, just to ensure that the data is opened to the outside world after the synchronization is completed, so the availability cannot be taken into account.Therefore, only partition fault tolerance and consistency can be satisfied.

Consistency pursued by ZooKeeper:

ZooKeeper does not pursue strong consistency when synchronizing data, but sequential consistency (monotonically increasing transaction id)