RedisPipelining底层原理分析及实践

见贤思齐

Wang FeiRedis是一种基于客户端-服务端模型以及请求/响应的TCP服务。在遇到批处理命令执行时，Redis提供了Pipelining(管道)来提升批处理性能。本文结合实践分析了Spring Boot框架下Redis的Lettuce客户端和Redisson客户端对Pipeline特性的支持原理，并针对实践过程中遇到的问题进行了分析，可以帮助开发者了解不同客户端对Pipeline支持原理及避免实际使用中出现问题。一、前言Redis 已经提供了像 mget 、mset 这种批量的命令，但是某些操作根本就不支持或没有批量的操作，从而与 Redis 高性能背道而驰。为此, Redis基于管道机制，提供Redis Pipeline新特性。Redis Pipeline是一种通过一次性发送多条命令并在执行完后一次性将结果返回，从而减少客户端与redis的通信次数来实现降低往返延时时间提升操作性能的技术。目前，Redis Pipeline是被很多个版本的Redis 客户端所支持的。二、Pipeline 底层原理分析2.1 Redis单个命令执行基本步骤Redis是一种基于客户端-服务端模型以及请求/响应的TCP服务。一次Redis客户端发起的请求，经过服务端的响应后，大致会经历如下的步骤：客户端发起一个（查询/插入）请求，并监听socket返回，通常情况都是阻塞模式等待Redis服务器的响应。服务端处理命令，并且返回处理结果给客户端。客户端接收到服务的返回结果，程序从阻塞代码处返回。2.2 RTT 时间Redis客户端和服务端之间通过网络连接进行数据传输，数据包从客户端到达服务器，并从服务器返回数据回复客户端的时间被称之为RTT(Round Trip Time - 往返时间)。我们可以很容易就意识到，Redis在连续请求服务端时，如果RTT时间为250ms, 即使Redis每秒能处理100k请求，但也会因为网络传输花费大量时间，导致每秒最多也只能处理4个请求，导致整体性能的下降。2.3 Redis Pipeline为了提升效率，这时候Pipeline出现了。Pipelining不仅仅能够降低RRT，实际上它极大的提升了单次执行的操作数。这是因为如果不使用Pipelining，那么每次执行单个命令,从访问数据的结构和服务端产生应答的角度，它的成本是很低的。但是从执行网络IO的角度，它的成本其实是很高的。其中涉及到read()和write()的系统调用，这意味着需要从用户态切换到内核态,而这个上下文的切换成本是巨大的。当使用Pipeline时，它允许多个命令的读通过一次read()操作，多个命令的应答使用一次write()操作，它允许客户端可以一次发送多条命令，而不等待上一条命令执行的结果。不仅减少了RTT，同时也减少了IO调用次数（IO调用涉及到用户态到内核态之间的切换），最终提升程序的执行效率与性能。如下图：要支持Pipeline，其实既要服务端的支持，也要客户端支持。对于服务端来说，所需要的是能够处理一个客户端通过同一个TCP连接发来的多个命令，可以理解为，这里将多个命令切分，和处理单个命令一样，Redis就是这样处理的。而客户端，则是要将多个命令缓存起来，缓冲区满了就发送，然后再写缓冲，最后才处理Redis的应答。三、Pipeline 基本使用及性能比较下面我们以给10w个set结构分别插入一个整数值为例，分别使用jedis单个命令插入、jedis使用Pipeline模式进行插入和redisson使用Pipeline模式进行插入以及测试其耗时。@Slf4jpublic class RedisPipelineTestDemo { public static void main(String[] args) { //连接redis Jedis jedis = new Jedis("10.101.17.180", 6379); //jedis逐一给每个set新增一个value String zSetKey = "Pipeline-test-set"; int size = 100000; long begin = System.currentTimeMillis(); for (int i = 0; i ids= Arrays.asList(1, 2, 3, 4, 5, 6, 7, 8, 9); Integer contentId = 1;redisTemplate.executePipelined(newInsertPipelineExecutionA(ids,contentId)); } @AllArgsConstructor private static class InsertPipelineExecutionA implements RedisCallback { private final List ids; private final Integer contentId; @Override public Void doInRedis(RedisConnection connection) DataAccessException { RedisSetCommands redisSetCommands = connection.setCommands(); ids.forEach(id-> { String redisKey = "aaa:" + id; String value = String.valueOf(contentId); redisSetCommands.sAdd(redisKey.getBytes(), value.getBytes()); }); return null; } }SessionCallback的使用方式：public void testSessionCallback() { List ids= Arrays.asList(1, 2, 3, 4, 5, 6, 7, 8, 9); Integer contentId = 1; redisTemplate.executePipelined(new InsertPipelineExecutionB(ids, contentId)); } @AllArgsConstructor private static class InsertPipelineExecutionB implements SessionCallback { private final List ids; private final Integer contentId; @Override public Void execute(RedisOperations operations) throws DataAccessException { SetOperations setOperations = (SetOperations) operations.opsForSet(); ids.forEach(id-> { String redisKey = "aaa:" + id; String value = String.valueOf(contentId); setOperations.add(redisKey, value); }); return null; } }4.3 RedisCallBack和SessionCallback之间的比较1、RedisCallBack和SessionCallback都可以实现回调，通过它们可以在同一条连接中一次执行多个redis命令。2、RedisCallback使用的是原生RedisConnection，用起来比较麻烦，比如上面执行set的add操作，key和value需要进行转换，可读性差，但原生api提供的功能比较齐全。3、SessionCalback提供了良好的封装，可以优先选择使用这种回调方式。最终的代码实现如下：public void executeB(List userIds, Integer iconId) { redisTemplate.executePipelined(new InsertPipelineExecution(userIds, iconId));} @AllArgsConstructorprivate static class InsertPipelineExecution implements SessionCallback { private final List userIds; private final Integer iconId; @Override public Void execute(RedisOperations operations) throws DataAccessException { SetOperations setOperations = (SetOperations) operations.opsForSet(); userIds.forEach(userId -> { String redisKey = "aaa:" + userId; String value = String.valueOf(iconId); setOperations.add(redisKey, value); }); return null; }}4.4 源码分析那么为什么使用Pipeline方式会对性能有较大提升呢，我们现在从源码入手着重分析一下：4.4.1 Pipeline方式下获取连接相关原理分析：@Override public List executePipelined(SessionCallback session, @Nullable RedisSerializer resultSerializer) { Assert.isTrue(initialized, "template not initialized; call afterPropertiesSet() before using it"); Assert.notNull(session, "Callback object must not be null"); //1. 获取对应的Redis连接工厂 RedisConnectionFactory factory = getRequiredConnectionFactory(); //2. 绑定连接过程 RedisConnectionUtils.bindConnection(factory, enableTransactionSupport); try { //3. 执行命令流程, 这里请求参数为RedisCallback, 里面有对应的回调操作 return execute((RedisCallback>) connection -> { //具体的回调逻辑 connection.openPipeline(); boolean PipelinedClosed = false; try { //执行命令 Object result = executeSession(session); if (result != null) { throw new InvalidDataAccessApiUsageException( "Callback cannot return a non-null value as it gets overwritten by the Pipeline"); } List closePipeline = connection.closePipeline(); PipelinedClosed = true; return deserializeMixedResults(closePipeline, resultSerializer, hashKeySerializer, hashValueSerializer); } finally { if (!PipelinedClosed) { connection.closePipeline(); } } }); } finally { RedisConnectionUtils.unbindConnection(factory); } }① 获取对应的Redis连接工厂，这里要使用Pipeline特性需要使用LettuceConnectionFactory方式，这里获取的连接工厂就是LettuceConnectionFactory。② 绑定连接过程，具体指的是将当前连接绑定到当前线程上面， 核心方法为：doGetConnection。public static RedisConnection doGetConnection(RedisConnectionFactory factory, boolean allowCreate, boolean bind, boolean enableTransactionSupport) { Assert.notNull(factory, "No RedisConnectionFactory specified"); //核心类，有缓存作用，下次可以从这里获取已经存在的连接 RedisConnectionHolder connHolder = (RedisConnectionHolder) TransactionSynchronizationManager.getResource(factory); //如果connHolder不为null, 则获取已经存在的连接, 提升性能 if (connHolder != null) { if (enableTransactionSupport) { potentiallyRegisterTransactionSynchronisation(connHolder, factory); } return connHolder.getConnection(); } ...... //第一次获取连接，需要从Redis连接工厂获取连接 RedisConnection conn = factory.getConnection(); //bind = true 执行绑定 if (bind) { RedisConnection connectionToBind = conn; ...... connHolder = new RedisConnectionHolder(connectionToBind); //绑定核心代码: 将获取的连接和当前线程绑定起来 TransactionSynchronizationManager.bindResource(factory, connHolder); ...... return connHolder.getConnection(); } return conn; }里面有个核心类RedisConnectionHolder，我们看一下RedisConnectionHolder connHolder =(RedisConnectionHolder)TransactionSynchronizationManager.getResource(factory);@Nullable public static Object getResource(Object key) { Object actualKey = TransactionSynchronizationUtils.unwrapResourceIfNecessary(key); Object value = doGetResource(actualKey); if (value != null & logger.isTraceEnabled()) { logger.trace("Retrieved value [" + value + "] for key [" + actualKey + "] bound to thread [" + Thread.currentThread().getName() + "]"); } return value; }里面有一个核心方法doGetResource(actualKey)，大家很容易猜测这里涉及到一个map结构，如果我们看源码，也确实是这样一个结构。@Nullable private static Object doGetResource(Object actualKey) { Map map = resources.get(); if (map == null) { return null; } Object value = map.get(actualKey); // Transparently remove ResourceHolder that was marked as void... if (value instanceof ResourceHolder & ((ResourceHolder) value).isVoid()) { map.remove(actualKey); // Remove entire ThreadLocal if empty... if (map.isEmpty()) { resources.remove(); } value = null; } return value; }resources是一个ThreadLocal类型，这里会涉及到根据RedisConnectionFactory获取到连接connection的逻辑,如果下一次是同一个actualKey，那么就直接使用已经存在的连接，而不需要新建一个连接。第一次这里map为null，就直接返回了，然后回到doGetConnection方法，由于这里bind为true，我们会执行TransactionSynchronizationManager.bindResource(factory, connHolder);，也就是将连接和当前线程绑定了起来。public static void bindResource(Object key, Object value) throws IllegalStateException { Object actualKey = TransactionSynchronizationUtils.unwrapResourceIfNecessary(key); Assert.notNull(value, "Value must not be null"); Map map = resources.get(); // set ThreadLocal Map if none found if (map == null) { map = new HashMap(); resources.set(map); } Object oldValue = map.put(actualKey, value); ...... }③我们回到executePipelined，在获取到连接工厂，将连接和当前线程绑定起来以后，就开始需要正式去执行命令了, 这里会调用execute方法@Override@Nullablepublic T execute(RedisCallback action) { return execute(action, isExposeConnection());}这里我们注意到execute方法的入参为RedisCallbackaction，RedisCallback对应的doInRedis操作如下,这里在后面的调用过程中会涉及到回调。connection.openPipeline();boolean PipelinedClosed = false;try { Object result = executeSession(session); if (result != null) { throw new InvalidDataAccessApiUsageException( "Callback cannot return a non-null value as it gets overwritten by the Pipeline"); } List closePipeline = connection.closePipeline(); PipelinedClosed = true; return deserializeMixedResults(closePipeline, resultSerializer, hashKeySerializer, hashValueSerializer);} finally { if (!PipelinedClosed) { connection.closePipeline(); }}我们再来看execute(action,isExposeConnection())方法，这里最终会调用execute(RedisCallbackaction, boolean exposeConnection, boolean Pipeline)方法。@Nullablepublic T execute(RedisCallback action, boolean exposeConnection, boolean Pipeline) { Assert.isTrue(initialized, "template not initialized; call afterPropertiesSet() before using it"); Assert.notNull(action, "Callback object must not be null"); //获取对应的连接工厂 RedisConnectionFactory factory = getRequiredConnectionFactory(); RedisConnection conn = null; try { if (enableTransactionSupport) { // only bind resources in case of potential transaction synchronization conn = RedisConnectionUtils.bindConnection(factory, enableTransactionSupport); } else { //获取对应的连接(enableTransactionSupport=false) conn = RedisConnectionUtils.getConnection(factory); } boolean existingConnection = TransactionSynchronizationManager.hasResource(factory); RedisConnection connToUse = preProcessConnection(conn, existingConnection); boolean PipelineStatus = connToUse.isPipelined(); if (Pipeline & !PipelineStatus) { connToUse.openPipeline(); } RedisConnection connToExpose = (exposeConnectionconnToUse : createRedisConnectionProxy(connToUse)); //核心方法，这里就开始执行回调操作 T result = action.doInRedis(connToExpose); // close Pipeline if (Pipeline & !PipelineStatus) { connToUse.closePipeline(); } // TODO: any other connection processing return postProcessResult(result, connToUse, existingConnection); } finally { RedisConnectionUtils.releaseConnection(conn, factory, enableTransactionSupport); }}我们看到这里最开始也是获取对应的连接工厂，然后获取对应的连接(enableTransactionSupport=false)，具体调用是RedisConnectionUtils.getConnection(factory)方法，最终会调用RedisConnection doGetConnection(RedisConnectionFactory factory, boolean allowCreate, boolean bind, boolean enableTransactionSupport)，此时bind为falsepublic static RedisConnection doGetConnection(RedisConnectionFactory factory, boolean allowCreate, boolean bind, boolean enableTransactionSupport) { Assert.notNull(factory, "No RedisConnectionFactory specified"); //直接获取与当前线程绑定的Redis连接 RedisConnectionHolder connHolder = (RedisConnectionHolder) TransactionSynchronizationManager.getResource(factory); if (connHolder != null) { if (enableTransactionSupport) { potentiallyRegisterTransactionSynchronisation(connHolder, factory); } return connHolder.getConnection(); } ...... return conn;}前面我们分析过一次，这里调用RedisConnectionHolder connHolder =(RedisConnectionHolder) TransactionSynchronizationManager.getResource(factory);会获取到之前和当前线程绑定的Redis，而不会新创建一个连接。然后会去执行T result = action.doInRedis(connToExpose)，这里的action为RedisCallback，执行doInRedis为：//开启Pipeline功能connection.openPipeline();boolean PipelinedClosed = false;try { //执行Redis命令 Object result = executeSession(session); if (result != null) { throw new InvalidDataAccessApiUsageException( "Callback cannot return a non-null value as it gets overwritten by the Pipeline"); } List closePipeline = connection.closePipeline(); PipelinedClosed = true; return deserializeMixedResults(closePipeline, resultSerializer, hashKeySerializer, hashValueSerializer);} finally { if (!PipelinedClosed) { connection.closePipeline(); }}这里最开始会开启Pipeline功能，然后执行Object result = executeSession(session);private Object executeSession(SessionCallback session) { return session.execute(this);}这里会调用我们自定义的execute方法@AllArgsConstructorprivate static class InsertPipelineExecution implements SessionCallback { private final List userIds; private final Integer iconId; @Override public Void execute(RedisOperations operations) throws DataAccessException { SetOperations setOperations = (SetOperations) operations.opsForSet(); userIds.forEach(userId -> { String redisKey = "aaa:" + userId; String value = String.valueOf(iconId); setOperations.add(redisKey, value); }); return null; }}进入到foreach循环，执行DefaultSetOperations的add方法。@Overridepublic Long add(K key, V... values) { byte[] rawKey = rawKey(key); byte[][] rawValues = rawValues((Object[]) values); //这里的connection.sAdd是后续回调要执行的方法 return execute(connection -> connection.sAdd(rawKey, rawValues), true);}这里会继续执行redisTemplate的execute方法，里面最终会调用我们之前分析过的T execute(RedisCallbackaction, boolean exposeConnection, boolean Pipeline)方法。@Nullablepublic T execute(RedisCallback action, boolean exposeConnection, boolean Pipeline) { Assert.isTrue(initialized, "template not initialized; call afterPropertiesSet() before using it"); Assert.notNull(action, "Callback object must not be null"); RedisConnectionFactory factory = getRequiredConnectionFactory(); RedisConnection conn = null; try { ...... //再次执行回调方法，这里执行的Redis基本数据结构对应的操作命令 T result = action.doInRedis(connToExpose); ...... // TODO: any other connection processing return postProcessResult(result, connToUse, existingConnection); } finally { RedisConnectionUtils.releaseConnection(conn, factory, enableTransactionSupport); }}这里会继续执行T result =action.doInRedis(connToExpose);，这里其实执行的doInRedis方法为：connection -> connection.sAdd(rawKey, rawValues)4.4.2 Pipeline方式下执行命令的流程分析:① 接着上面的流程分析，这里的sAdd方法实际调用的是DefaultStringRedisConnection的sAdd方法@Overridepublic Long sAdd(byte[] key, byte[]... values) { return convertAndReturn(delegate.sAdd(key, values), identityConverter);}②这里会进一步调用DefaultedRedisConnection的sAdd方法@Override@Deprecateddefault Long sAdd(byte[] key, byte[]... values) { return setCommands().sAdd(key, values);}③接着调用LettuceSetCommands的sAdd方法@Overridepublic Long sAdd(byte[] key, byte[]... values) { Assert.notNull(key, "Key must not be null!"); Assert.notNull(values, "Values must not be null!"); Assert.noNullElements(values, "Values must not contain null elements!"); try { // 如果开启了 Pipelined 模式，获取的是 异步连接，进行异步操作 if (isPipelined()) { Pipeline(connection.newLettuceResult(getAsyncConnection().sadd(key, values))); return null; } if (isQueueing()) { transaction(connection.newLettuceResult(getAsyncConnection().sadd(key, values))); return null; } //常规模式下，使用的是同步操作 return getConnection().sadd(key, values); } catch (Exception ex) { throw convertLettuceAccessException(ex); }}这里我们开启了Pipeline, 实际会调用Pipeline(connection.newLettuceResult(getAsyncConnection().sadd(key, values))); 也就是获取异步连接getAsyncConnection，然后进行异步操作sadd，而常规模式下，使用的是同步操作，所以在Pipeline模式下，执行效率更高。从上面的获取连接和具体命令执行相关源码分析可以得出使用Lettuce客户端Pipeline模式高效的根本原因：普通模式下，每执行一个命令都需要先打开一个连接，命令执行完毕以后又需要关闭这个连接，执行下一个命令时，又需要经过连接打开和关闭的流程；而Pipeline的所有命令的执行只需要经过一次连接打开和关闭。普通模式下命令的执行是同步阻塞模式，而Pipeline模式下命令的执行是异步非阻塞模式。五、项目中遇到的坑前面介绍了涉及到批量操作，可以使用Redis Pipelining机制，那是不是任何批量操作相关的场景都可以使用呢，比如list类型数据的批量移除操作，我们的代码最开始是这么写的：public void deleteSet(String updateKey, Set userIds) { if (CollectionUtils.isEmpty(userIds)) { return; } redisTemplate.executePipelined(new DeleteListCallBack(userIds, updateKey)); } @AllArgsConstructorprivate static class DeleteListCallBack implements SessionCallback { private Set userIds; private String updateKey; @Override public Object execute(RedisOperations operations) throws DataAccessException { ListOperations listOperations = (ListOperations) operations.opsForList(); userIds.forEach(userId -> listOperations.remove(updateKey, 1, userId.toString())); return null; }}在数据量比较小的时候没有出现问题，直到有一条收到了Redis的内存和cpu利用率的告警消息，我们发现这么使用是有问题的，核心原因在于list的lrem操作的时间复杂度是O(N+M)，其中N是list的长度, M是要移除的元素的个数，而我们这里还是一个一个移除的，当然会导致Redis数据积压和cpu每秒ops升高导致cpu利用率飚高。也就是说，即使使用Pipeline进行批量操作，但是由于单次操作很耗时，是会导致整个Redis出现问题的。后面我们进行了优化，选用了list的ltrim命令，一次命令执行批量remove操作：public void deleteSet(String updateKey, Set deviceIds) { if (CollectionUtils.isEmpty(deviceIds)) { return; } int maxSize = 10000; redisTemplate.opsForList().trim(updateKey, maxSize + 1, -1); }由于ltrim本身的时间复杂度为O(M), 其中M要移除的元素的个数，相比于原始方案的lrem，效率提升很多，可以不需要使用Redis Pipeline，优化结果使得Redis内存利用率和cpu利用率都极大程度得到缓解。六、Redisson 对 Redis Pipeline 特性支持在redisson官方文档中额外特性介绍中有说到批量命令执行这个特性， 也就是多个命令在一次网络调用中集中发送，该特性是RBatch这个类支持的，从这个类的描述来看，主要是为Redis Pipeline这个特性服务的，并且主要是通过队列和异步实现的。/** * Interface for using Redis Pipeline feature. * * All method invocations on objects got through this interface * are batched to separate queue and could be executed later * with execute() or executeAsync() methods. * * * @author Nikita Koksharov * */public interface RBatch { /** * Returns stream instance by name * * @param type of key * @param type of value * @param name of stream * @return RStream object */ RStreamAsync getStream(String name); /** * Returns stream instance by name * using provided codec for entries. * * @param type of key * @param type of value * @param name - name of stream * @param codec - codec for entry * @return RStream object */ RStreamAsync getStream(String name, Codec codec); ...... /** * Returns list instance by name. * * @param type of object * @param name - name of object * @return List object */ RListAsync getList(String name); RListAsync getList(String name, Codec codec); ...... /** * Executes all operations accumulated during async methods invocations. * * If cluster configuration used then operations are grouped by slot ids * and may be executed on different servers. Thus command execution order could be changed * * @return List with result object for each command * @throws RedisException in case of any error * */ BatchResult execute() throws RedisException; /** * Executes all operations accumulated during async methods invocations asynchronously. * * In cluster configurations operations grouped by slot ids * so may be executed on different servers. Thus command execution order could be changed * * @return List with result object for each command */ RFuture> executeAsync(); /** * Discard batched commands and release allocated buffers used for parameters encoding. */ void discard(); /** * Discard batched commands and release allocated buffers used for parameters encoding. * * @return void */ RFuture discardAsync(); }简单的测试代码如下：@Slf4jpublic class RedisPipelineTest { public static void main(String[] args) { //Redisson使用Pipeline模式 Config config = new Config(); config.useSingleServer().setAddress("redis://xx.xx.xx.xx:6379"); RedissonClient redisson = Redisson.create(config); RBatch redisBatch = redisson.createBatch(); int size = 100000; String zSetKey = "Pipeline-test-set"; long begin = System.currentTimeMillis(); //将命令放入队列中 for (int i = 0; i addAsync(V e) { String name = getRawName(e); return commandExecutor.writeAsync(name, codec, RedisCommands.SADD_SINGLE, name, encode(e));}（1）接着调用CommandAsyncExecutor的异步写入方法writeAsync。@Overridepublic RFuture writeAsync(String key, Codec codec, RedisCommand command, Object... params) { RPromise mainPromise = createPromise(); NodeSource source = getNodeSource(key); async(false, source, codec, command, params, mainPromise, false); return mainPromise;}（2）接着调用批量命令执行器CommandBatchService的异步发送命令。@Overridepublic void async(boolean readOnlyMode, NodeSource nodeSource, Codec codec, RedisCommand command, Object[] params, RPromise mainPromise, boolean ignoreRedirect) { if (isRedisBasedQueue()) { boolean isReadOnly = options.getExecutionMode() == ExecutionMode.REDIS_READ_ATOMIC; RedisExecutor executor = new RedisQueuedBatchExecutor(isReadOnly, nodeSource, codec, command, params, mainPromise, false, connectionManager, objectBuilder, commands, connections, options, index, executed, latch, referenceType); executor.execute(); } else { //执行分支 RedisExecutor executor = new RedisBatchExecutor(readOnlyMode, nodeSource, codec, command, params, mainPromise, false, connectionManager, objectBuilder, commands, options, index, executed, referenceType); executor.execute(); } }（3）接着调用了RedisBatchExecutor.execute方法和BaseRedisBatchExecutor.addBatchCommandData方法。@Overridepublic void execute() { addBatchCommandData(params);} protected final void addBatchCommandData(Object[] batchParams) { MasterSlaveEntry msEntry = getEntry(source); Entry entry = commands.get(msEntry); if (entry == null) { entry = new Entry(); Entry oldEntry = commands.putIfAbsent(msEntry, entry); if (oldEntry != null) { entry = oldEntry; } } if (!readOnlyMode) { entry.setReadOnlyMode(false); } Codec codecToUse = getCodec(codec); BatchCommandData commandData = new BatchCommandData(mainPromise, codecToUse, command, batchParams, index.incrementAndGet()); entry.getCommands().add(commandData);}这里的commands以主节点为KEY,以待发送命令队列列表为VALUE(Entry)，保存一个MAP.然后会把命令都添加到entry的commands命令队列中， Entry结构如下面代码所示。public static class Entry { Deque> commands = new LinkedBlockingDeque(); volatile boolean readOnlyMode = true; public Deque> getCommands() { return commands; } public void setReadOnlyMode(boolean readOnlyMode) { this.readOnlyMode = readOnlyMode; } public boolean isReadOnlyMode() { return readOnlyMode; } public void clearErrors() { for (BatchCommandData commandEntry : commands) { commandEntry.clearError(); } } }4. 批量执行命令redisBatch.execute()，这里会最终调用CommandBatchService的executeAsync方法，该方法完整代码如下，我们下面来逐一进行拆解。public RFuture> executeAsync() { ...... RPromise> promise = new RedissonPromise(); RPromise voidPromise = new RedissonPromise(); if (this.options.isSkipResult() & this.options.getSyncSlaves() == 0) { ...... } else { //这里是对异步执行结果进行处理,可以先忽略, 后面会详细讲，先关注批量执行命令的逻辑 voidPromise.onComplete((res, ex) -> { ...... }); } AtomicInteger slots = new AtomicInteger(commands.size()); ...... //真正执行的代码入口，批量执行命令 for (Map.Entry e : commands.entrySet()) { RedisCommonBatchExecutor executor = new RedisCommonBatchExecutor(new NodeSource(e.getKey()), voidPromise, connectionManager, this.options, e.getValue(), slots, referenceType); executor.execute(); } return promise; }里面会用到我们在3.3步骤所生成的commands实例。（1）接着调用了基类RedisExecutor的execute方法public void execute() { ...... connectionFuture.onComplete((connection, e) -> { if (connectionFuture.isCancelled()) { connectionManager.getShutdownLatch().release(); return; } if (!connectionFuture.isSuccess()) { connectionManager.getShutdownLatch().release(); exception = convertException(connectionFuture); return; } //调用RedisCommonBatchExecutor的sendCommand方法, 里面会将多个命令放到一个List> list列表里面 sendCommand(attemptPromise, connection); writeFuture.addListener(new ChannelFutureListener() { @Override public void operationComplete(ChannelFuture future) throws Exception { checkWriteFuture(writeFuture, attemptPromise, connection); } }); }); ...... }（2）接着调用RedisCommonBatchExecutor的sendCommand方法，里面会将多个命令放到一个List list列表里面。@Override protected void sendCommand(RPromise attemptPromise, RedisConnection connection) { boolean isAtomic = options.getExecutionMode() != ExecutionMode.IN_MEMORY; boolean isQueued = options.getExecutionMode() == ExecutionMode.REDIS_READ_ATOMIC || options.getExecutionMode() == ExecutionMode.REDIS_WRITE_ATOMIC; //将多个命令放到一个List> list列表里面 List> list = new ArrayList(entry.getCommands().size()); if (source.getRedirect() == Redirect.ASK) { RPromise promise = new RedissonPromise(); list.add(new CommandData(promise, StringCodec.INSTANCE, RedisCommands.ASKING, new Object[] {})); } for (CommandData c : entry.getCommands()) { if ((c.getPromise().isCancelled() || c.getPromise().isSuccess()) & !isWaitCommand(c) & !isAtomic) { // skip command continue; } list.add(c); } ...... //调用RedisConnection的send方法，将命令一次性发到Redis服务器端 writeFuture = connection.send(new CommandsData(attemptPromise, list, options.isSkipResult(), isAtomic, isQueued, options.getSyncSlaves() > 0)); }（3）接着调用RedisConnection的send方法，通过Netty通信发送命令到Redis服务器端执行，这里也验证了Redisson客户端底层是采用Netty进行通信的。public ChannelFuture send(CommandsData data) { return channel.writeAndFlush(data);}5. 接收返回结果，这里主要是监听事件是否完成，然后组装返回结果, 核心方法是步骤4提到的CommandBatchService的executeAsync方法,里面会对返回结果进行监听和处理, 核心代码如下：public RFuture> executeAsync() { ...... RPromise> promise = new RedissonPromise(); RPromise voidPromise = new RedissonPromise(); if (this.options.isSkipResult() & this.options.getSyncSlaves() == 0) { ...... } else { voidPromise.onComplete((res, ex) -> { //对返回结果的处理 executed.set(true); ...... List responses = new ArrayList(entries.size()); int syncedSlaves = 0; for (BatchCommandData commandEntry : entries) { if (isWaitCommand(commandEntry)) { syncedSlaves = (Integer) commandEntry.getPromise().getNow(); } else if (!commandEntry.getCommand().getName().equals(RedisCommands.MULTI.getName()) & !commandEntry.getCommand().getName().equals(RedisCommands.EXEC.getName()) & !this.options.isSkipResult()) { ...... //获取单个命令的执行结果 Object entryResult = commandEntry.getPromise().getNow(); ...... //将单个命令执行结果放到List中 responses.add(entryResult); } } BatchResult result = new BatchResult(responses, syncedSlaves); promise.trySuccess(result); ...... }); } ...... return promise;}这里会把单个命令的执行结果放到responses里面，最终返回RPromisepromise。从上面的分析来看，Redisson客户端对Redis Pipeline的支持也是从多个命令在一次网络通信中执行和异步处理来实现的。七、总结Redis提供了Pipelining进行批量操作的高级特性，极大地提高了部分数据类型没有批量执行命令导致的执行耗时而引起的性能问题，但是我们在使用的过程中需要考虑Pipeline操作中单个命令执行的耗时问题，否则带来的效果可能适得其反。最后扩展分析了Redisson客户端对Redis Pipeline特性的支持原理，可以与Lettuce客户端对Redis Pipeline支持原理进行比较，加深Pipeline在不同Redis客户端实现方式的理解。参考资料：Redis PipeliningRedisTemplate使用Pipeline管道命令如何使用好Redis PipelineRedisson 管道批量发送命令流程分析END猜你喜欢MySQL 主从 AUTO_INCREMENT 不一致问题分析vivo 消息中间件测试环境项目多版本实践Java 8 内存管理原理解析及内存故障排查实践