Redis cluster deployment and application scenarios

summary

redis cluster People like to call it redis colony , Some people like to call it partitioned cluster . Sum up ,redis There are three patterns , Stand alone , Sentinel mode , Fragmentation cluster . On the surface, more and more machines are used , In fact, it does change with the amount of data read and write , Three deployment modes need to be upgraded continuously , Let's look at the difference , And applicable scenarios .

Schematic diagram of partitioned cluster

explain ： Pictured , The meaning of partitioned cluster is clear at a glance , Data is distributed and stored on different nodes . First, the client will respond to key Conduct CRC16 Calculation , Calculate to get a slot Number , Then the data is read and written in the corresponding redis For instance . Compare with sentinel mode , Sharding cluster not only ensures high availability , At the same time, it can also store data in pieces , It really achieves the function of sharing the reading and writing pressure .

Fragmented cluster application scenarios

1. When the storage data is large , A single machine certainly won't work . It is easy to trigger the elimination strategy frequently , Frequent cache misses , Cause avalanches , breakdown , through , meanwhile 
  RDB fork Child processes tend to block the main thread 

2. When the storage data is large , Sentinel mode doesn't work either ( You can read the pressure separately ). In addition to the problems of the single machine , It will also cause inconsistency between master and slave synchronization data , Ring buffer