Java Redisson 操作 Redis 指南

最后修改:2024年8月22日

by baeldung

1. Overview

Redisson 是一个用于 Java 的 Redis client。In this article, we’ll explore some of its features, and demonstrate how it could facilitate building distributed business applications.

Redisson constitutes an in-memory data grid that offers distributed Java objects and services backed by Redis**.** Its distributed in-memory data model allows sharing of domain objects and services across applications and servers.

In this article, we’ll see how to set up Redisson, understand how it operates, and explore some of Redisson’s objects and services.

2. Maven 依赖

添加 Redisson Maven 依赖：

<dependency>
    <groupId>org.redisson</groupId>
    <artifactId>redisson</artifactId>
    <version>3.20.0</version>
</dependency>

最新版本可以查看 Maven 中央仓库.

3. 配置

Before we get started, we must ensure we have the latest version of Redis set up and running. If you don’t have Redis and you use Linux or Macintosh, you can follow the information here to get it set up. If you’re a Windows user, you can set up Redis using this unofficial post which is archived since February 2021.

Redisson 支持多种连接配置：

  • 单节点
  • 主从模式
  • 哨兵模式
  • 集群
  • 副本

Redisson supports Amazon Web Services (AWS) ElastiCache Cluster and Azure Redis Cache for Clustered and Replicated Nodes.

Let’s connect to a single node instance of Redis. This instance is running locally on the default port, 6379:

RedissonClient client = Redisson.create();

You can pass different configurations to the Redisson object’s create method. This could be configurations to have it connect to a different port, or maybe, to connect to a Redis cluster. This configuration could be in Java code or loaded from an external configuration file.

3.1. Java 配置

Let’s configure Redisson in Java code:

Config config = new Config();
config.useSingleServer()
  .setAddress("redis://127.0.0.1:6379");

RedissonClient client = Redisson.create(config);

We specify Redisson configurations in an instance of a Config object and then pass it to the create method. Above, we specified to Redisson that we want to connect to a single node instance of Redis. To do this we used the Config object’s useSingleServer method. This returns a reference to a SingleServerConfig object.

The SingleServerConfig object has settings that Redisson uses to connect to a single node instance of Redis. Here, we use its setAddress method to configure the address setting. This sets the address of the node we’re connecting to. Some other settings include retryAttempts, connectionTimeout and clientName. These settings are configured using their corresponding setter methods.

We can configure Redisson for different Redis configurations in a similar way using the Config object’s following methods:

  • useSingleServer – for single node instance. Get single node settings here
  • useMasterSlaveServers – for master with slave nodes. Get master-slave node settings here
  • useSentinelServers – for sentinel nodes. Get sentinel node settings here
  • useClusterServers – for clustered nodes. Get clustered node settings here
  • useReplicatedServers – for replicated nodes. Get replicated node settings here

3.2. 配置文件方式

Redisson can load configurations from external JSON or YAML files:

Config config = Config.fromJSON(new File("singleNodeConfig.json"));  
RedissonClient client = Redisson.create(config);

The Config object’s fromJSON method can load configurations from a string, file, input stream or URL.

Here is the sample configuration in the singleNodeConfig.json file:

{
    "singleServerConfig": {
        "idleConnectionTimeout": 10000,
        "connectTimeout": 10000,
        "timeout": 3000,
        "retryAttempts": 3,
        "retryInterval": 1500,
        "password": null,
        "subscriptionsPerConnection": 5,
        "clientName": null,
        "address": "redis://127.0.0.1:6379",
        "subscriptionConnectionMinimumIdleSize": 1,
        "subscriptionConnectionPoolSize": 50,
        "connectionMinimumIdleSize": 10,
        "connectionPoolSize": 64,
        "database": 0,
        "dnsMonitoringInterval": 5000
    },
    "threads": 0,
    "nettyThreads": 0,
    "codec": null
}

对应的 YAML 配置文件:

singleServerConfig:
    idleConnectionTimeout: 10000
    connectTimeout: 10000
    timeout: 3000
    retryAttempts: 3
    retryInterval: 1500
    password: null
    subscriptionsPerConnection: 5
    clientName: null
    address: "redis://127.0.0.1:6379"
    subscriptionConnectionMinimumIdleSize: 1
    subscriptionConnectionPoolSize: 50
    connectionMinimumIdleSize: 10
    connectionPoolSize: 64
    database: 0
    dnsMonitoringInterval: 5000
threads: 0
nettyThreads: 0
codec: !<org.redisson.codec.JsonJacksonCodec> {}

We can configure other Redis configurations from a file in a similar manner using settings peculiar to that configuration. For your reference, here are their JSON and YAML file formats:

To save a Java configuration to JSON or YAML format, we can use the toJSON or toYAML methods of the Config object:

Config config = new Config();
// ... we configure multiple settings here in Java
String jsonFormat = config.toJSON();
String yamlFormat = config.toYAML();

Now that we know how to configure Redisson, let’s look at how Redisson executes operations.

4. 操作

Redisson supports synchronous, asynchronous and reactive interfaces. Operations over these interfaces are thread-safe.

All entities (objects, collections, locks and services) generated by a RedissonClient have synchronous and asynchronous methods. Synchronous methods bear asynchronous variants. These methods normally bear the same method name of their synchronous variants appended with “Async”. Let’s look at a synchronous method of the RAtomicLong object:

RedissonClient client = Redisson.create();
RAtomicLong myLong = client.getAtomicLong('myLong');

The asynchronous variant of the synchronous compareAndSet method would be:

RFuture<Boolean> isSet = myLong.compareAndSetAsync(6, 27);

The asynchronous variant of the method returns an RFuture object. We can set listeners on this object to get back the result when it becomes available:

isSet.handle((result, exception) -> {
    // handle the result or exception here.
});

To generate reactive objects, we would need to use the RedissonReactiveClient:

RedissonReactiveClient client = Redisson.createReactive();
RAtomicLongReactive myLong = client.getAtomicLong("myLong");

Publisher<Boolean> isSetPublisher = myLong.compareAndSet(5, 28);

This method returns reactive objects based on the Reactive Streams Standard for Java 9.

Let’s explore some of the distributed objects provided by Redisson.

5. 对象

An individual instance of a Redisson object is serialized and stored in any of the available Redis nodes backing Redisson. These objects could be distributed in a cluster across multiple nodes and can be accessed by a single application or multiple applications/servers.

These distributed objects follow specifications from the java.util.concurrent.atomic package. They support lock-free, thread-safe and atomic operations on objects stored in Redis. Data consistency between applications/servers is ensured as values are not updated while another application is reading the object.

Redisson objects are bound to Redis keys. We can manage these keys through the RKeys interface. And then, we access our Redisson objects using these keys.

There are several options we may use to get the Redis keys.

We can simply get all the keys:

RKeys keys = client.getKeys();

Alternatively, we can extract only the names:

Iterable<String> allKeys = keys.getKeys();

And finally, we’re able to get the keys conforming to a pattern:

Iterable<String> keysByPattern = keys.getKeysByPattern('key*')

The RKeys interface also allows deleting keys, deleting keys by pattern and other useful key-based operations that we could use to manage our keys and objects.

Redisson 提供下面几种分布式对象：

  • ObjectHolder
  • BinaryStreamHolder
  • GeospatialHolder
  • BitSet
  • AtomicLong
  • AtomicDouble
  • Topic
  • BloomFilter
  • HyperLogLog

Let’s take a look at three of these objects: ObjectHolder, AtomicLong, and Topic.

5.1. Object Holder

Represented by the RBucket class, this object can hold any type of object. This object has a maximum size of 512MB:

RBucket<Ledger> bucket = client.getBucket("ledger");
bucket.set(new Ledger());
Ledger ledger = bucket.get();

The RBucket object can perform atomic operations such as compareAndSet and getAndSet on objects it holds.

5.2. AtomicLong

Represented by the RAtomicLong class, this object closely resembles the java.util.concurrent.atomic.AtomicLong class and represents a long value that can be updated atomically:

RAtomicLong atomicLong = client.getAtomicLong("myAtomicLong");
atomicLong.set(5);
atomicLong.incrementAndGet();

5.3. Topic

The Topic object supports Redis’ “publish and subscribe” mechanism. To listen for published messages:

RTopic subscribeTopic = client.getTopic("baeldung");
subscribeTopic.addListener(CustomMessage.class,
  (channel, customMessage) -> future.complete(customMessage.getMessage()));

Above, the Topic is registered to listen to messages from the “baeldung” channel. We then add a listener to the topic to handle incoming messages from that channel. We can add multiple listeners to a channel.

Let’s publish messages to the “baeldung” channel:

RTopic publishTopic = client.getTopic("baeldung");
long clientsReceivedMessage
  = publishTopic.publish(new CustomMessage("This is a message"));

This could be published from another application or server. The CustomMessage object will be received by the listener and processed as defined in the onMessage method.

We can learn more about other Redisson objects here.

6. 集合

We handle Redisson collections in the same fashion we handle objects.

Redisson 提供的分布式集合：

  • Map
  • Multimap
  • Set
  • SortedSet
  • ScoredSortedSet
  • LexSortedSet
  • List
  • Queue
  • Deque
  • BlockingQueue
  • BoundedBlockingQueue
  • BlockingDeque
  • BlockingFairQueue
  • DelayedQueue
  • PriorityQueue
  • PriorityDeque

Let’s take a look at three of these collections: Map, Set, and List.

6.1. Map

Redisson-based maps implement the java.util.concurrent.ConcurrentMap and java.util.Map interfaces. Redisson has four map implementations. These are RMap, RMapCache, RLocalCachedMap and RClusteredMap.

Let’s create a map with Redisson:

RMap<String, Ledger> map = client.getMap("ledger");
Ledger newLedger = map.put("123", new Ledger());map

RMapCache supports map entry eviction. RLocalCachedMap allows local caching of map entries*. RClusteredMap* allows data from a single map to be split across Redis cluster master nodes.

We can learn more about Redisson maps here.

6.2. Set

Redisson-based Set implements the java.util.Set interface.

Redisson has three Set implementations, RSet, RSetCache, and RClusteredSet with similar functionality as their map counterparts.

Let’s create a Set with Redisson:

RSet<Ledger> ledgerSet = client.getSet("ledgerSet");
ledgerSet.add(new Ledger());

We can learn more about Redisson sets here.

6.3. List

Redisson-based Lists implement the java.util.List interface.

Let’s create a List with Redisson:

RList<Ledger> ledgerList = client.getList("ledgerList");
ledgerList.add(new Ledger());

We can learn more about other Redisson collections here.

7. 锁和线程同步

Redisson’s distributed locks allow for thread synchronization across applications/servers. Redisson’s list of locks and synchronizers includes:

  • Lock
  • FairLock
  • MultiLock
  • ReadWriteLock
  • Semaphore
  • PermitExpirableSemaphore
  • CountDownLatch

Let’s take a look at Lock and MultiLock.

7.1. Lock

Redisson’s Lock implements java.util.concurrent.locks.Lock interface.

Let’s implement a lock, represented by the RLock class:

RLock lock = client.getLock("lock");
lock.lock();
// perform some long operations...
lock.unlock();

7.2. MultiLock

Redisson’s RedissonMultiLock groups multiple RLock objects and treats them as a single lock:

RLock lock1 = clientInstance1.getLock("lock1");
RLock lock2 = clientInstance2.getLock("lock2");
RLock lock3 = clientInstance3.getLock("lock3");

RedissonMultiLock lock = new RedissonMultiLock(lock1, lock2, lock3);
lock.lock();
// perform long running operation...
lock.unlock();

We can learn more about other locks here.

8. Services

Redisson exposes 4 types of distributed services. These are Remote Service, Live Object Service, Executor Service and Scheduled Executor Service. Let’s look at the Remote Service and Live Object Service.

8.1. Remote Service

This service provides Java remote method invocation facilitated by Redis. A Redisson remote service consists of a server-side (worker instance) and client-side implementation. The server-side implementation executes a remote method invoked by the client. Calls from a remote service can be synchronous or asynchronous.

The server side registers an interface for remote invocation:

RRemoteService remoteService = client.getRemoteService();
LedgerServiceImpl ledgerServiceImpl = new LedgerServiceImpl();

remoteService.register(LedgerServiceInterface.class, ledgerServiceImpl);

The client side calls a method of the registered remote interface:

RRemoteService remoteService = client.getRemoteService();
LedgerServiceInterface ledgerService
  = remoteService.get(LedgerServiceInterface.class);

List<String> entries = ledgerService.getEntries(10);

We can learn more about remote services here.

8.2. Live Object Service

Redisson Live Objects extend the concept of standard Java objects that could only be accessed from a single JVM to enhanced Java objects that could be shared between different JVMs in different machines. This is accomplished by mapping an object’s fields to a Redis hash. This mapping is made through a runtime-constructed proxy class. Field getters and setters are mapped to Redis hget/hset commands.

Redisson Live Objects support atomic field access as a result of Redis’ single-threaded nature.

Creating a Live Object is simple:

@REntity
public class LedgerLiveObject {
    @RId
    private String name;

    // getters and setters...
}

We annotate our class with @REntity and a unique or identifying field with @RId. Once we have done this, we can use our Live Object in our application:

RLiveObjectService service = client.getLiveObjectService();

LedgerLiveObject ledger = new LedgerLiveObject();
ledger.setName("ledger1");

ledger = service.persist(ledger);

We create our Live Object like standard Java objects using the new keyword. We then use an instance of RLiveObjectService to save the object to Redis using its persist method.

If the object has previously been persisted to Redis, we can retrieve the object:

LedgerLiveObject returnLedger
  = service.get(LedgerLiveObject.class, "ledger1");

We use the RLiveObjectService to get our Live Object using the field annotated with @RId.

Here we can find more about Redisson Live Objects, and other Redisson services are described here.

9. Pipelining

Redisson supports pipelining. Multiple operations can be batched as a single atomic operation. This is facilitated by the RBatch class. Multiple commands are aggregated against an RBatch object instance before they are executed:

RBatch batch = client.createBatch();
batch.getMap("ledgerMap").fastPutAsync("1", "2");
batch.getMap("ledgerMap").putAsync("2", "5");

BatchResult<?> batchResult = batch.execute();

10. Scripting

Redisson supports LUA scripting. We can execute LUA scripts against Redis:

client.getBucket("foo").set("bar");
String result = client.getScript().eval(Mode.READ_ONLY,
  "return redis.call('get', 'foo')", RScript.ReturnType.VALUE);

11. Low-Level Client

It’s possible that we might want to perform Redis operations not yet supported by Redisson. Redisson provides a low-level client that allows the execution of native Redis commands:

RedisClientConfig redisClientConfig = new RedisClientConfig();
redisClientConfig.setAddress("localhost", 6379);

RedisClient client = RedisClient.create(redisClientConfig);

RedisConnection conn = client.connect();
conn.sync(StringCodec.INSTANCE, RedisCommands.SET, "test", 0);

conn.closeAsync();
client.shutdown();

The low-level client also supports asynchronous operations.

12. 总结

This article showcased Redisson and some of the features that make it ideal for developing distributed applications. We explored its distributed objects, collections, locks and services. We also explored some of its other features such as pipelining, scripting and its low-level client.

Redisson also provides integration with other frameworks such as the JCache API, Spring Cache, Hibernate Cache and Spring Sessions. We can learn more about its integration with other frameworks here.

You can find code samples in the GitHub project.

« 上一篇： Java 9中CompletableFuture API的改进
» 下一篇： Google Guice介绍