Project practice, redis cluster technology learning (VII)

Redis.4 Cluster scaling

Redis.4.1 Stretching principle

Redis Cluster provides flexible node expansion and contraction scheme . Without affecting the external services of the cluster

Next , You can add nodes to the cluster to expand capacity, or you can downsize some nodes offline , Pictured

Redis The cluster can realize flexible online and offline control of nodes . The principle can be abstracted as the flexible movement of slots and corresponding data between different nodes . First, let's look at the cluster slot we built before and the corresponding relationship between data and nodes , As shown in the figure

The three master nodes maintain their own slots and corresponding data , If you want to join 1 When a node realizes cluster expansion , The data slot needs to be related to the new node through the command , Pictured

Each node in the figure migrates part of the slot and data to a new node 6385, Each node is responsible for fewer slots and data than before, so as to achieve the purpose of cluster expansion . Here, we deliberately ignore the details of slot and data migration between nodes , The purpose is to let readers focus on the upper slot and node allocation , Understand the upper principle of horizontal scaling of clusters ： Cluster scaling = The movement of slots and data between nodes , Next, we will introduce the details of cluster expansion and contraction .

10.4.2 Expand clusters

Capacity expansion is the most common requirement of distributed storage ,Redis Cluster capacity expansion can be divided into the following steps ：

1） Prepare new nodes .

2） To join the cluster .

3） Migration slots and data .

1. Prepare new nodes

New nodes need to be prepared in advance and run in cluster mode , The new node is recommended to be configured according to the nodes in the cluster

bring into correspondence with , Easy to manage and unify . After the configuration is ready, start the two nodes with the following commands ：

redis-server conf/redis-6385.conf

redis-server conf/redis-6386.conf

The new node after startup runs as an orphan node , No other node communicates with it , The cluster structure is shown in the figure

Shown .

2. To join the cluster

The new node still adopts cluster meet Command to join an existing cluster . Execute on any node in the cluster

cluster meet Order to let 6385 and 6386 Nodes join , The order is as follows ：

127.0.0.1:6379> cluster meet 127.0.0.1 6385

127.0.0.1:6379> cluster meet 127.0.0.1 6386

After the new node is added, the cluster structure is shown in the figure .

The old and new nodes in the cluster have changed over a period of time ping/pong After message communication , All nodes will send

Create new nodes and save their state locally . For example, we are in 6380 Execution on node cluster

nodes The command can see the new node information ：

127.0.0.1:6380>cluster ndoes

1a205dd8b2819a00dd1e8b6be40a8e2abe77b756 127.0.0.1:6385 master - 0

1469347800759

7 connected

475528b1bcf8e74d227104a6cf1bf70f00c24aae 127.0.0.1:6386 master - 0

1469347798743

8 connected

...

At the beginning, new nodes are all primary nodes , But because there is no responsible slot , So I can't accept any

How to read and write . For the subsequent operations of new nodes, we generally have two options ：

· Expand its migration slot and data .

· The slave node as the other master node is responsible for the failover .

redis-trib.rb The tool also implements the command to add new nodes to an existing cluster , It also realizes direct addition

Add support for slave nodes , The order is as follows ：

redis-trib.rb add-node 127.0.0.1:6385 127.0.0.1:6379

redis-trib.rb add-node 127.0.0.1:6386 127.0.0.1:6379

（Redis5.0 After the version, also put redis-trib.rb The order was changed to redis-cli ）