Redis cluster

Redis colony

Concept

Not enough capacity ,redis How to expand ？

1. Changing one set to multiple sets can solve colony

Concurrent write operations , redis How to apportion ？

1. colony

in addition , A master-slave mode , Pass on mode , Host down , Lead to ip The address has changed , The corresponding host address needs to be modified for configuration in the application 、 Port and other information .

Previously, it was solved by proxy host , however redis3.0 The solution . That is, decentralized cluster configuration .

Proxy host ： A proxy is used between the client and the server It is different from the agent Redis The server sends the request receive data

shortcoming ： There are many servers Inconvenient maintenance

No center :

Request to call the cluster , Automatically transfer to the target server

3 Lord 3 from

Redis The cluster has realized to Redis Horizontal expansion of , Start now N individual redis node , The entire database is distributed and stored here N A node in the , Each node stores the total data 1/N.

Redis Clusters are partitioned （partition） To provide a certain degree of usability （availability）： Even if some nodes in the cluster fail or fail to communicate , The cluster can also continue to process command requests .

To configure

Turn on daemonize yes

Pid File name

Designated port

Log File name

Dump.rdb name

Appendonly Turn it off or change the name

In profile ：

cluster-enabled yes Open cluster mode

cluster-config-file nodes-6379.conf Set the node profile name

cluster-node-timeout 15000 Set the node loss time , Beyond that time （ millisecond ）, The cluster automatically switches between master and slave .

With 6 For example 6 Ports simulation 6 Servers

3678 Port of , Other ports are the same as the following, just change the number

start-up 6 individual redis Before service composition , Please make sure that all redis After the instance is started ,nodes-xxxx.conf All files are generated normally

The new version （6.2） There is no need to install ruby Environmental Science

Go to start installation redis Of src Next

Execute... Under this path

redis-cli --cluster create --cluster-replicas 1 192.168.200.130:6379 192.168.200.130:6380 192.168.200.130:6381 192.168.200.130:6389 192.168.200.130:6390 192.168.200.130:6391

6 A cluster of nodes

• –cluste Indicates cluster operation
• –cluster-replicas The way of clustering 1 It means the simplest way to configure the cluster , A mainframe , One slave , At present, there are exactly three groups .2 Namely A mainframe , Two slaves

Don't use it here 127.0.0.1, Please use real IP Address

Master and slave have been assigned . Accept or not

[OK] All 16384 slots covered.

redis-cli -c -p 6379 -c Use cluster policy to connect , The setting data will be automatically switched to the corresponding write host

cluster nodes View cluster information

What is? slots

Cluster building completed There's a hint

One Redis The cluster contains 16384 Slots （hash slot）, Every key in the database belongs here 16384 One of the slots ,

The cluster uses the formula CRC16(key) % 16384 To calculate the key key Which slot does it belong to , among CRC16(key) Statements are used to evaluate keys key Of CRC16 The checksum .

Each node in the cluster is responsible for processing a portion of the slots . for instance , If a cluster can have a master node , among ：

node A Responsible for handling 0 No. to 5460 Slot number .

node B Responsible for handling 5461 No. to 10922 Slot number .set

node C Responsible for handling 10923 No. to 16383 Slot number .

Not in one slot Key values under , Can't use mget,mset Wait for multi key operation .

Multiple values added Braces are required . Can pass {} To define the concept of a group , So that key in {} Key value pairs with the same content in a slot In the middle .

mset <key1><value1><key2><value2>

 mset name{user} aa age{user} 20 

CLUSTER keyslot X Calculation x Slot value for

CLUSTER countkeysinslot <slot> Calculate the number of keys in this card slot （ Only those within the corresponding range of the current library , If it's another library , Returns the 0）

CLUSTER GETKEYSINSLOT <slot><count> return count individual slot Key in groove .

Fault recovery

• If the master node goes offline G 了 The slave node is automatically promoted to the master node Be careful ：15 Second timeout

• When the master node comes back, it will become a slave .

• If the master and slave of a certain slot （6 Hang up 2 individual ） All hang up , and cluster-require-full-coverage by yes , that , The whole cluster is down

  If the master and slave of a certain slot hang up , and cluster-require-full-coverage by no , that , None of the slot data is available , Can't store .

Jedis Related operations of

Even if the host is not connected , The cluster automatically switches host storage . Host write , Read from machine .

Decentralized master-slave cluster . No matter which host the data is written from , Data can be read on other hosts .

public class JedisClusterTest {
    
  public static void main(String[] args) {
    
// Set<HostAndPort> set =new HashSet<HostAndPort>();
// //HostAndPort  Any one will do   Each can be used as the entrance of the cluster 
// set.add();
     JedisCluster jedisCluster=new JedisCluster(new HostAndPort("192.168.200.130",6379)); // Pass in set Or a HostAndPort  Fine 
     jedisCluster.set("bbb1", "v111");
     System.out.println(jedisCluster.get("bbb1"));
  }
}

Advantages and disadvantages

Redis Clustering provides the following benefits

• Realize capacity expansion

• Apportionment pressure

• No central configuration is relatively simple

Redis The lack of clustering

• Multi key operation is not supported

• Multibond Redis Transactions are not supported .lua Scripts are not supported

• Due to the late emergence of the cluster scheme , Many companies have adopted other clustering schemes , The proxy or client partition scheme wants to migrate to redis cluster, Overall migration is required rather than gradual transition , High complexity .