Cap principle

CAP Principles are also called CAP Theorem , In a distributed system , Consistency（ Uniformity ）、 Availability（ Usability ）、Partition tolerance（ Partition tolerance ）, You can't have all three .
Uniformity （C）： All data backup in distributed system , Is it the same value at the same time .（ Equivalent to all nodes accessing the same copy of the latest data ）
Usability （A）： After some nodes in the cluster fail , Whether the cluster as a whole can respond to the client's read and write requests .（ High availability for data updates ）
Zone tolerance （P）： In terms of actual effect , Partition is equivalent to the time limit requirement for communication . If the system fails to reach data consistency within the time limit , It means that the partition happened , The current operation must be in C and A Make a choice between .
CAP The essence of the principle is either AP, or CP, or AC, But it doesn't exist CAP. If there is no copy of the data in a distributed system , Then the system must satisfy the strong consistency condition , Because there's only one data , There will be no data inconsistency , here C and P There are two elements , But if the system has network partition status or downtime , Inevitably, some data cannot be accessed , At this point, the availability condition cannot be satisfied , In this case, we get CP System , however CAP Can't satisfy at the same time .
So when designing distributed architecture , Must make a choice . At present, the performance of the system is generally improved by the final consistency of each node in the distributed cache , Cluster data consistency is realized by using data asynchronous replication technology between multiple nodes . Usually use something like memcached And so on. NOSQL As a means of implementation . although memcached It can also be a distributed cluster environment , But for a piece of data , It's always stored on a certain computer memcached Server . In case of network failure or server crash , All data stored on this server will be inaccessible . Because the data is stored in memory , Restart the server , Will result in all data loss . Of course, you can also implement a set of mechanisms by yourself , Used in distributed systems memcached Data synchronization and persistence between , But the implementation is very difficult .