Ribbon 是netflix 公司開(kāi)源的基于客戶(hù)端的負(fù)載均衡組件，是Spring Cloud大家庭中非常重要的一個(gè)模塊找都；Ribbon應(yīng)該也是整個(gè)大家庭中相對(duì)而言比較復(fù)雜的模塊，直接影響到服務(wù)調(diào)度的質(zhì)量和性能廊敌。全面掌握Ribbon可以幫助我們了解在分布式微服務(wù)集群工作模式下飒炎，服務(wù)調(diào)度應(yīng)該考慮到的每個(gè)環(huán)節(jié)谁不。
本文將詳細(xì)地剖析Ribbon的設(shè)計(jì)原理，幫助大家對(duì)Spring Cloud 有一個(gè)更好的認(rèn)知震嫉。

一. Spring集成下的Ribbon工作結(jié)構(gòu)

先貼一張總覽圖森瘪，說(shuō)明一下Spring如何集成Ribbon的，如下所示：

image.png

Spring Cloud集成模式下的Ribbon有以下幾個(gè)特征：

1. Ribbon 服務(wù)配置方式
   每一個(gè)服務(wù)配置都有一個(gè)Spring ApplicationContext上下文票堵，用于加載各自服務(wù)的實(shí)例扼睬。
   比如，當(dāng)前Spring Cloud 系統(tǒng)內(nèi)悴势，有如下幾個(gè)服務(wù)：

mobile-bff服務(wù)在實(shí)際使用中窗宇，會(huì)用到order和user模塊，那么在mobile-bff服務(wù)的Spring上下文中特纤，會(huì)為order 和user 分別創(chuàng)建一個(gè)子ApplicationContext,用于加載各自服務(wù)模塊的配置军俊。也就是說(shuō)，各個(gè)客戶(hù)端的配置相互獨(dú)立捧存，彼此不收影響

2. 和Feign的集成模式
   在使用Feign作為客戶(hù)端時(shí)粪躬，最終請(qǐng)求會(huì)轉(zhuǎn)發(fā)成 http://<服務(wù)名稱(chēng)>/<relative-path-to-service>的格式，通過(guò)LoadBalancerFeignClient昔穴， 提取出服務(wù)標(biāo)識(shí)<服務(wù)名稱(chēng)>镰官，然后根據(jù)服務(wù)名稱(chēng)在上下文中查找對(duì)應(yīng)服務(wù)的負(fù)載均衡器FeignLoadBalancer，負(fù)載均衡器負(fù)責(zé)根據(jù)既有的服務(wù)實(shí)例的統(tǒng)計(jì)信息吗货，挑選出最合適的服務(wù)實(shí)例

二泳唠、Spring Cloud模式下和Feign的集成實(shí)現(xiàn)方式

和Feign結(jié)合的場(chǎng)景下，F(xiàn)eign的調(diào)用會(huì)被包裝成調(diào)用請(qǐng)求LoadBalancerCommand宙搬，然后底層通過(guò)Rxjava基于事件的編碼風(fēng)格笨腥，發(fā)送請(qǐng)求孙援；Spring Cloud框架通過(guò) Feigin 請(qǐng)求的URL，提取出服務(wù)名稱(chēng)扇雕，然后在上下文中找到對(duì)應(yīng)服務(wù)的的負(fù)載均衡器實(shí)現(xiàn)FeignLoadBalancer,然后通過(guò)負(fù)載均衡器中挑選一個(gè)合適的Server實(shí)例,然后將調(diào)用請(qǐng)求轉(zhuǎn)發(fā)到該Server實(shí)例上拓售，完成調(diào)用，在此過(guò)程中镶奉，記錄對(duì)應(yīng)Server實(shí)例的調(diào)用統(tǒng)計(jì)信息础淤。

/**
     * Create an {@link Observable} that once subscribed execute network call asynchronously with a server chosen by load balancer.
     * If there are any errors that are indicated as retriable by the {@link RetryHandler}, they will be consumed internally by the
     * function and will not be observed by the {@link Observer} subscribed to the returned {@link Observable}. If number of retries has
     * exceeds the maximal allowed, a final error will be emitted by the returned {@link Observable}. Otherwise, the first successful
     * result during execution and retries will be emitted.
     */
    public Observable<T> submit(final ServerOperation<T> operation) {
        final ExecutionInfoContext context = new ExecutionInfoContext();
        
        if (listenerInvoker != null) {
            try {
                listenerInvoker.onExecutionStart();
            } catch (AbortExecutionException e) {
                return Observable.error(e);
            }
        }
        
        // 同一Server最大嘗試次數(shù)
        final int maxRetrysSame = retryHandler.getMaxRetriesOnSameServer();
        //下一Server最大嘗試次數(shù)
        final int maxRetrysNext = retryHandler.getMaxRetriesOnNextServer();

        // Use the load balancer
        // 使用負(fù)載均衡器，挑選出合適的Server哨苛，然后執(zhí)行Server請(qǐng)求鸽凶，將請(qǐng)求的數(shù)據(jù)和行為整合到ServerStats中
        Observable<T> o = 
                (server == null ? selectServer() : Observable.just(server))
                .concatMap(new Func1<Server, Observable<T>>() {
                    @Override
                    // Called for each server being selected
                    public Observable<T> call(Server server) {
                        // 獲取Server的統(tǒng)計(jì)值
                        context.setServer(server);
                        final ServerStats stats = loadBalancerContext.getServerStats(server);
                        
                        // Called for each attempt and retry 服務(wù)調(diào)用
                        Observable<T> o = Observable
                                .just(server)
                                .concatMap(new Func1<Server, Observable<T>>() {
                                    @Override
                                    public Observable<T> call(final Server server) {
                                        context.incAttemptCount();//重試計(jì)數(shù)
                                        loadBalancerContext.noteOpenConnection(stats);//鏈接統(tǒng)計(jì)
                                        
                                        if (listenerInvoker != null) {
                                            try {
                                                listenerInvoker.onStartWithServer(context.toExecutionInfo());
                                            } catch (AbortExecutionException e) {
                                                return Observable.error(e);
                                            }
                                        }
                                        //執(zhí)行監(jiān)控器，記錄執(zhí)行時(shí)間
                                        final Stopwatch tracer = loadBalancerContext.getExecuteTracer().start();
                                        //找到合適的server后建峭，開(kāi)始執(zhí)行請(qǐng)求
                                        //底層調(diào)用有結(jié)果后玻侥，做消息處理
                                        return operation.call(server).doOnEach(new Observer<T>() {
                                            private T entity;
                                            @Override
                                            public void onCompleted() {
                                                recordStats(tracer, stats, entity, null);
                                                // 記錄統(tǒng)計(jì)信息
                                            }

                                            @Override
                                            public void onError(Throwable e) {
                                                recordStats(tracer, stats, null, e);//記錄異常信息
                                                logger.debug("Got error {} when executed on server {}", e, server);
                                                if (listenerInvoker != null) {
                                                    listenerInvoker.onExceptionWithServer(e, context.toExecutionInfo());
                                                }
                                            }

                                            @Override
                                            public void onNext(T entity) {
                                                this.entity = entity;//返回結(jié)果值
                                                if (listenerInvoker != null) {
                                                    listenerInvoker.onExecutionSuccess(entity, context.toExecutionInfo());
                                                }
                                            }                            
                                            
                                            private void recordStats(Stopwatch tracer, ServerStats stats, Object entity, Throwable exception) {
                                                tracer.stop();//結(jié)束計(jì)時(shí)
                                                //標(biāo)記請(qǐng)求結(jié)束，更新統(tǒng)計(jì)信息
                                                loadBalancerContext.noteRequestCompletion(stats, entity, exception, tracer.getDuration(TimeUnit.MILLISECONDS), retryHandler);
                                            }
                                        });
                                    }
                                });
                        //如果失敗亿蒸，根據(jù)重試策略觸發(fā)重試邏輯
                        // 使用observable 做重試邏輯凑兰，根據(jù)predicate 做邏輯判斷，這里做
                        if (maxRetrysSame > 0) 
                            o = o.retry(retryPolicy(maxRetrysSame, true));
                        return o;
                    }
                });
         // next請(qǐng)求處理边锁，基于重試器操作   
        if (maxRetrysNext > 0 && server == null) 
            o = o.retry(retryPolicy(maxRetrysNext, false));
        
        return o.onErrorResumeNext(new Func1<Throwable, Observable<T>>() {
            @Override
            public Observable<T> call(Throwable e) {
                if (context.getAttemptCount() > 0) {
                    if (maxRetrysNext > 0 && context.getServerAttemptCount() == (maxRetrysNext + 1)) {
                        e = new ClientException(ClientException.ErrorType.NUMBEROF_RETRIES_NEXTSERVER_EXCEEDED,
                                "Number of retries on next server exceeded max " + maxRetrysNext
                                + " retries, while making a call for: " + context.getServer(), e);
                    }
                    else if (maxRetrysSame > 0 && context.getAttemptCount() == (maxRetrysSame + 1)) {
                        e = new ClientException(ClientException.ErrorType.NUMBEROF_RETRIES_EXEEDED,
                                "Number of retries exceeded max " + maxRetrysSame
                                + " retries, while making a call for: " + context.getServer(), e);
                    }
                }
                if (listenerInvoker != null) {
                    listenerInvoker.onExecutionFailed(e, context.toFinalExecutionInfo());
                }
                return Observable.error(e);
            }
        });
    }

從一組ServerList 列表中挑選合適的Server

    /**
     * Compute the final URI from a partial URI in the request. The following steps are performed:
     * <ul>
     * <li>  如果host尚未指定姑食，則從負(fù)載均衡器中選定 host/port
     * <li>  如果host 尚未指定并且尚未找到負(fù)載均衡器，則嘗試從 虛擬地址中確定host/port
     * <li> 如果指定了HOST,并且URI的授權(quán)部分通過(guò)虛擬地址設(shè)置茅坛，并且存在負(fù)載均衡器音半，則通過(guò)負(fù)載就均衡器中確定host/port（指定的HOST將會(huì)被忽略）
     * <li> 如果host已指定，但是尚未指定負(fù)載均衡器和虛擬地址配置贡蓖，則使用真實(shí)地址作為host
     * <li> if host is missing but none of the above applies, throws ClientException
     * </ul>
     *
     * @param original Original URI passed from caller
     */
    public Server getServerFromLoadBalancer(@Nullable URI original, @Nullable Object loadBalancerKey) throws ClientException {
        String host = null;
        int port = -1;
        if (original != null) {
            host = original.getHost();
        }
        if (original != null) {
            Pair<String, Integer> schemeAndPort = deriveSchemeAndPortFromPartialUri(original);        
            port = schemeAndPort.second();
        }

        // Various Supported Cases
        // The loadbalancer to use and the instances it has is based on how it was registered
        // In each of these cases, the client might come in using Full Url or Partial URL
        ILoadBalancer lb = getLoadBalancer();
        if (host == null) {
            // 提供部分URI曹鸠，缺少HOST情況下
            // well we have to just get the right instances from lb - or we fall back
            if (lb != null){
                Server svc = lb.chooseServer(loadBalancerKey);// 使用負(fù)載均衡器選擇Server
                if (svc == null){
                    throw new ClientException(ClientException.ErrorType.GENERAL,
                            "Load balancer does not have available server for client: "
                                    + clientName);
                }
                //通過(guò)負(fù)載均衡器選擇的結(jié)果中選擇host
                host = svc.getHost();
                if (host == null){
                    throw new ClientException(ClientException.ErrorType.GENERAL,
                            "Invalid Server for :" + svc);
                }
                logger.debug("{} using LB returned Server: {} for request {}", new Object[]{clientName, svc, original});
                return svc;
            } else {
                // No Full URL - and we dont have a LoadBalancer registered to
                // obtain a server
                // if we have a vipAddress that came with the registration, we
                // can use that else we
                // bail out
                // 通過(guò)虛擬地址配置解析出host配置返回
                if (vipAddresses != null && vipAddresses.contains(",")) {
                    throw new ClientException(
                            ClientException.ErrorType.GENERAL,
                            "Method is invoked for client " + clientName + " with partial URI of ("
                            + original
                            + ") with no load balancer configured."
                            + " Also, there are multiple vipAddresses and hence no vip address can be chosen"
                            + " to complete this partial uri");
                } else if (vipAddresses != null) {
                    try {
                        Pair<String,Integer> hostAndPort = deriveHostAndPortFromVipAddress(vipAddresses);
                        host = hostAndPort.first();
                        port = hostAndPort.second();
                    } catch (URISyntaxException e) {
                        throw new ClientException(
                                ClientException.ErrorType.GENERAL,
                                "Method is invoked for client " + clientName + " with partial URI of ("
                                + original
                                + ") with no load balancer configured. "
                                + " Also, the configured/registered vipAddress is unparseable (to determine host and port)");
                    }
                } else {
                    throw new ClientException(
                            ClientException.ErrorType.GENERAL,
                            this.clientName
                            + " has no LoadBalancer registered and passed in a partial URL request (with no host:port)."
                            + " Also has no vipAddress registered");
                }
            }
        } else {
            // Full URL Case URL中指定了全地址，可能是虛擬地址或者是hostAndPort
            // This could either be a vipAddress or a hostAndPort or a real DNS
            // if vipAddress or hostAndPort, we just have to consult the loadbalancer
            // but if it does not return a server, we should just proceed anyways
            // and assume its a DNS
            // For restClients registered using a vipAddress AND executing a request
            // by passing in the full URL (including host and port), we should only
            // consult lb IFF the URL passed is registered as vipAddress in Discovery
            boolean shouldInterpretAsVip = false;

            if (lb != null) {
                shouldInterpretAsVip = isVipRecognized(original.getAuthority());
            }
            if (shouldInterpretAsVip) {
                Server svc = lb.chooseServer(loadBalancerKey);
                if (svc != null){
                    host = svc.getHost();
                    if (host == null){
                        throw new ClientException(ClientException.ErrorType.GENERAL,
                                "Invalid Server for :" + svc);
                    }
                    logger.debug("using LB returned Server: {} for request: {}", svc, original);
                    return svc;
                } else {
                    // just fall back as real DNS
                    logger.debug("{}:{} assumed to be a valid VIP address or exists in the DNS", host, port);
                }
            } else {
                // consult LB to obtain vipAddress backed instance given full URL
                //Full URL execute request - where url!=vipAddress
                logger.debug("Using full URL passed in by caller (not using load balancer): {}", original);
            }
        }
        // end of creating final URL
        if (host == null){
            throw new ClientException(ClientException.ErrorType.GENERAL,"Request contains no HOST to talk to");
        }
        // just verify that at this point we have a full URL

        return new Server(host, port);
    }

三. LoadBalancer--負(fù)載均衡器的核心

LoadBalancer 的職能主要有三個(gè)：

1. 維護(hù)Sever列表的數(shù)量(新增斥铺、更新彻桃、刪除等)
2. 維護(hù)Server列表的狀態(tài)(狀態(tài)更新)
3. 當(dāng)請(qǐng)求Server實(shí)例時(shí)，能否返回最合適的Server實(shí)例

本章節(jié)將通過(guò)詳細(xì)闡述著這三個(gè)方面仅父。

3.1 負(fù)載均衡器的內(nèi)部基本實(shí)現(xiàn)原理

先熟悉一下負(fù)載均衡器LoadBalancer的實(shí)現(xiàn)原理圖：

image.png

image.png

3.2 如何維護(hù)Server列表他膳？(新增、更新绒窑、刪除)

單從服務(wù)列表的維護(hù)角度上棕孙，Ribbon的結(jié)構(gòu)如下所示：

image.png

Server列表的維護(hù)從實(shí)現(xiàn)方法上分為兩類(lèi):

1. 基于配置的服務(wù)列表
   這種方式一般是通過(guò)配置文件，靜態(tài)地配置服務(wù)器列表些膨，這種方式相對(duì)而言比較簡(jiǎn)單蟀俊，但并不是意味著在機(jī)器運(yùn)行的時(shí)候就一直不變。netflix 在做Spring cloud 套件時(shí)傀蓉，使用了分布式配置框架netflix archaius 欧漱，archaius 框架有一個(gè)特點(diǎn)是會(huì)動(dòng)態(tài)的監(jiān)控配置文件的變化职抡，將變化刷新到各個(gè)應(yīng)用上葬燎。也就是說(shuō)，當(dāng)我們?cè)诓魂P(guān)閉服務(wù)的情況下缚甩，如果修改了基于配置的服務(wù)列表時(shí), 服務(wù)列表可以直接刷新
2. 結(jié)合服務(wù)發(fā)現(xiàn)組件（如Eureka）的服務(wù)注冊(cè)信息動(dòng)態(tài)維護(hù)服務(wù)列表
   基于Spring Cloud框架下谱净，服務(wù)注冊(cè)和發(fā)現(xiàn)是一個(gè)分布式服務(wù)集群必不可少的一個(gè)組件，它負(fù)責(zé)維護(hù)不同的服務(wù)實(shí)例（注冊(cè)擅威、續(xù)約壕探、取消注冊(cè)），本文將介紹和Eureka集成模式下郊丛，如果借助Eureka的服務(wù)注冊(cè)信息動(dòng)態(tài)刷新ribbon 的服務(wù)列表

Ribbon 通過(guò)配置項(xiàng)：<service-name>.ribbon.NIWSServerListClassName 來(lái)決定使用哪種實(shí)現(xiàn)方式李请。對(duì)應(yīng)地：

Server列表可能在運(yùn)行的時(shí)候動(dòng)態(tài)的更新，而具體的更新方式由<service-name>.ribbon.ServerListUpdaterClassName,當(dāng)前有如下兩種實(shí)現(xiàn)方式：

• 基于定時(shí)任務(wù)拉取服務(wù)列表方式
  這種方式的實(shí)現(xiàn)為：com.netflix.loadbalancer.PollingServerListUpdater,其內(nèi)部維護(hù)了一個(gè)周期性的定時(shí)任務(wù)厉熟，拉取最新的服務(wù)列表导盅，然后將最新的服務(wù)列表更新到ServerList之中，其核心的實(shí)現(xiàn)邏輯如下：

public class PollingServerListUpdater implements ServerListUpdater {
    private static final Logger logger = LoggerFactory.getLogger(PollingServerListUpdater.class);

    private static long LISTOFSERVERS_CACHE_UPDATE_DELAY = 1000; // msecs;
    private static int LISTOFSERVERS_CACHE_REPEAT_INTERVAL = 30 * 1000; // msecs;
    // 更新器線程池定義以及鉤子設(shè)置
    private static class LazyHolder {
        private final static String CORE_THREAD = "DynamicServerListLoadBalancer.ThreadPoolSize";
        private final static DynamicIntProperty poolSizeProp = new DynamicIntProperty(CORE_THREAD, 2);
        private static Thread _shutdownThread;

        static ScheduledThreadPoolExecutor _serverListRefreshExecutor = null;

        static {
            int coreSize = poolSizeProp.get();
            ThreadFactory factory = (new ThreadFactoryBuilder())
                    .setNameFormat("PollingServerListUpdater-%d")
                    .setDaemon(true)
                    .build();
            _serverListRefreshExecutor = new ScheduledThreadPoolExecutor(coreSize, factory);
            poolSizeProp.addCallback(new Runnable() {
                @Override
                public void run() {
                    _serverListRefreshExecutor.setCorePoolSize(poolSizeProp.get());
                }

            });
            _shutdownThread = new Thread(new Runnable() {
                public void run() {
                    logger.info("Shutting down the Executor Pool for PollingServerListUpdater");
                    shutdownExecutorPool();
                }
            });
            Runtime.getRuntime().addShutdownHook(_shutdownThread);
        }

        private static void shutdownExecutorPool() {
            if (_serverListRefreshExecutor != null) {
                _serverListRefreshExecutor.shutdown();

                if (_shutdownThread != null) {
                    try {
                        Runtime.getRuntime().removeShutdownHook(_shutdownThread);
                    } catch (IllegalStateException ise) { // NOPMD
                        // this can happen if we're in the middle of a real
                        // shutdown,
                        // and that's 'ok'
                    }
                }

            }
        }
    }
    // 省略部分代碼
    @Override
    public synchronized void start(final UpdateAction updateAction) {
        if (isActive.compareAndSet(false, true)) {
            //創(chuàng)建定時(shí)任務(wù),按照特定的實(shí)行周期執(zhí)行更新操作
            final Runnable wrapperRunnable = new Runnable() {
                @Override
                public void run() {
                    if (!isActive.get()) {
                        if (scheduledFuture != null) {
                            scheduledFuture.cancel(true);
                        }
                        return;
                    }
                    try {
                        //執(zhí)行update操作 揍瑟，更新操作定義在LoadBalancer中
                        updateAction.doUpdate();
                        lastUpdated = System.currentTimeMillis();
                    } catch (Exception e) {
                        logger.warn("Failed one update cycle", e);
                    }
                }
            };
           //定時(shí)任務(wù)創(chuàng)建
            scheduledFuture = getRefreshExecutor().scheduleWithFixedDelay(
                    wrapperRunnable,
                    initialDelayMs, //初始延遲時(shí)間
                    refreshIntervalMs, //內(nèi)部刷新時(shí)間
                    TimeUnit.MILLISECONDS
            );
        } else {
            logger.info("Already active, no-op");
        }
    }
//省略部分代碼
}

有上述代碼可以看到,ServerListUpdator只是定義了更新的方式白翻，而具體怎么更新，則是封裝成UpdateAction來(lái)操作的：

    /**
     * an interface for the updateAction that actually executes a server list update
     */
    public interface UpdateAction {
        void doUpdate();
    }
    
    //在DynamicServerListLoadBalancer 中則實(shí)現(xiàn)了具體的操作：
    public DynamicServerListLoadBalancer() {
        this.isSecure = false;
        this.useTunnel = false;
        this.serverListUpdateInProgress = new AtomicBoolean(false);
        this.updateAction = new UpdateAction() {
            public void doUpdate() {
                //更新服務(wù)列表
                DynamicServerListLoadBalancer.this.updateListOfServers();
            }
        };
    }

    @VisibleForTesting
    public void updateListOfServers() {
        List<T> servers = new ArrayList();
       // 通過(guò)ServerList獲取最新的服務(wù)列表 
       if (this.serverListImpl != null) {
            servers = this.serverListImpl.getUpdatedListOfServers();
            LOGGER.debug("List of Servers for {} obtained from Discovery client: {}", this.getIdentifier(), servers);
            //返回的結(jié)果通過(guò)過(guò)濾器的方式進(jìn)行過(guò)濾
            if (this.filter != null) {
                servers = this.filter.getFilteredListOfServers((List)servers);
                LOGGER.debug("Filtered List of Servers for {} obtained from Discovery client: {}", this.getIdentifier(), servers);
            }
        }
        //更新列表
        this.updateAllServerList((List)servers);
    }

    protected void updateAllServerList(List<T> ls) {
        if (this.serverListUpdateInProgress.compareAndSet(false, true)) {
            try {
                Iterator var2 = ls.iterator();

                while(var2.hasNext()) {
                    T s = (Server)var2.next();
                    s.setAlive(true);
                }

                this.setServersList(ls);
                super.forceQuickPing();
            } finally {
                this.serverListUpdateInProgress.set(false);
            }
        }

    }

• 基于Eureka服務(wù)事件通知的方式更新
  基于Eureka的更新方式則有些不同, 當(dāng)Eureka注冊(cè)中心發(fā)生了Server服務(wù)注冊(cè)信息變更時(shí)绢片，會(huì)將消息通知發(fā)送到EurekaNotificationServerListUpdater 上滤馍，然后此Updator觸發(fā)刷新ServerList：

public class EurekaNotificationServerListUpdater implements ServerListUpdater {

   //省略部分代碼
  
    @Override
    public synchronized void start(final UpdateAction updateAction) {
        if (isActive.compareAndSet(false, true)) {
          //創(chuàng)建Eureka時(shí)間監(jiān)聽(tīng)器岛琼，當(dāng)Eureka發(fā)生改變后，將觸發(fā)對(duì)應(yīng)邏輯  
          this.updateListener = new EurekaEventListener() {
                @Override
                public void onEvent(EurekaEvent event) {
                    if (event instanceof CacheRefreshedEvent) {
                        //內(nèi)部消息隊(duì)列
                       if (!updateQueued.compareAndSet(false, true)) {  // if an update is already queued
                            logger.info("an update action is already queued, returning as no-op");
                            return;
                        }

                        if (!refreshExecutor.isShutdown()) {
                            try {
                                //提交更新操作請(qǐng)求到消息隊(duì)列中
                                refreshExecutor.submit(new Runnable() {
                                    @Override
                                    public void run() {
                                        try {
                                            updateAction.doUpdate(); // 執(zhí)行真正的更新操作
                                            lastUpdated.set(System.currentTimeMillis());
                                        } catch (Exception e) {
                                            logger.warn("Failed to update serverList", e);
                                        } finally {
                                            updateQueued.set(false);
                                        }
                                    }
                                });  // fire and forget
                            } catch (Exception e) {
                                logger.warn("Error submitting update task to executor, skipping one round of updates", e);
                                updateQueued.set(false);  // if submit fails, need to reset updateQueued to false
                            }
                        }
                        else {
                            logger.debug("stopping EurekaNotificationServerListUpdater, as refreshExecutor has been shut down");
                            stop();
                        }
                    }
                }
            };
            //EurekaClient 客戶(hù)端實(shí)例
            if (eurekaClient == null) {
                eurekaClient = eurekaClientProvider.get();
            }
            //基于EeurekaClient注冊(cè)事件監(jiān)聽(tīng)器
            if (eurekaClient != null) {
                eurekaClient.registerEventListener(updateListener);
            } else {
                logger.error("Failed to register an updateListener to eureka client, eureka client is null");
                throw new IllegalStateException("Failed to start the updater, unable to register the update listener due to eureka client being null.");
            }
        } else {
            logger.info("Update listener already registered, no-op");
        }
    }

}

3.2.1 相關(guān)的配置項(xiàng)

3.2.2 ribbon的默認(rèn)實(shí)現(xiàn)

ribbon在默認(rèn)情況下阁苞，會(huì)采用如下的配置項(xiàng)随珠，即，采用基于配置的服務(wù)列表維護(hù)猬错，基于定時(shí)任務(wù)按時(shí)拉取服務(wù)列表的方式窗看，頻率為30s.

<service-name>.ribbon.NIWSServerListClassName=com.netflix.loadbalancer.ConfigurationBasedServerList
<service-name>.ribbon.listOfServers=<ip:port>,<ip:port>
<service-name>.ribbon.ServerListUpdaterClassName=com.netflix.loadbalancer.EurekaNotificationServerListUpdater
<service-name>.ribbon.ServerListRefreshInterval=30
### 更新線程池大小
DynamicServerListLoadBalancer.ThreadPoolSize=2

3.2.3 Spring Cloud集成下的配置

ribbon在默認(rèn)情況下，會(huì)采用如下的配置項(xiàng)倦炒，即显沈，采用基于配置的服務(wù)列表維護(hù)，基于定時(shí)任務(wù)按時(shí)拉取服務(wù)列表的方式逢唤，頻率為30s.

<service-name>.ribbon.NIWSServerListClassName=com.netflix.loadbalancer.DiscoveryEnabledNIWSServerList
<service-name>.ribbon.ServerListUpdaterClassName=com.netflix.loadbalancer.EurekaNotificationServerListUpdater
### 更新線程池大小
EurekaNotificationServerListUpdater.ThreadPoolSize=2
###通知隊(duì)列接收大小
EurekaNotificationServerListUpdater.queueSize=1000

3.3 負(fù)載均衡器如何維護(hù)服務(wù)實(shí)例的狀態(tài)拉讯？

Ribbon負(fù)載均衡器將服務(wù)實(shí)例的狀態(tài)維護(hù)托交給Pinger、 PingerStrategy鳖藕、IPing 來(lái)維護(hù)魔慷，具體交互模式如下所示：

image.png

/**
* 定義Ping服務(wù)狀態(tài)是否有效的策略，是序列化順序Ping著恩，還是并行的方式Ping院尔，在此過(guò)程中，應(yīng)當(dāng)保證相互不受影響
 *
 */
public interface IPingStrategy {

    boolean[] pingServers(IPing ping, Server[] servers);
}


/**
 * 定義如何Ping一個(gè)服務(wù)，確保是否有效
 * @author stonse
 *
 */
public interface IPing {
    
    /**
     * Checks whether the given <code>Server</code> is "alive" i.e. should be
     * considered a candidate while loadbalancing
     * 校驗(yàn)是否存活
     */
    public boolean isAlive(Server server);
}

3.3.1 創(chuàng)建Ping定時(shí)任務(wù)

默認(rèn)情況下，負(fù)載均衡器內(nèi)部會(huì)創(chuàng)建一個(gè)周期性定時(shí)任務(wù)

默認(rèn)的PingStrategy仇穗，采用序列化的實(shí)現(xiàn)方式绳瘟，依次檢查服務(wù)實(shí)例是否可用:

/**
     * Default implementation for <c>IPingStrategy</c>, performs ping
     * serially, which may not be desirable, if your <c>IPing</c>
     * implementation is slow, or you have large number of servers.
     */
    private static class SerialPingStrategy implements IPingStrategy {

        @Override
        public boolean[] pingServers(IPing ping, Server[] servers) {
            int numCandidates = servers.length;
            boolean[] results = new boolean[numCandidates];

            logger.debug("LoadBalancer:  PingTask executing [{}] servers configured", numCandidates);

            for (int i = 0; i < numCandidates; i++) {
                results[i] = false; /* Default answer is DEAD. */
                try {
                    // 按序列依次檢查服務(wù)是否正常，并返回對(duì)應(yīng)的數(shù)組表示 
                    if (ping != null) {
                        results[i] = ping.isAlive(servers[i]);
                    }
                } catch (Exception e) {
                    logger.error("Exception while pinging Server: '{}'", servers[i], e);
                }
            }
            return results;
        }
    }

3.3.2 Ribbon默認(rèn)的IPing實(shí)現(xiàn):DummyPing

默認(rèn)的IPing實(shí)現(xiàn),直接返回為true:

public class DummyPing extends AbstractLoadBalancerPing {

    public DummyPing() {
    }

    public boolean isAlive(Server server) {
        return true;
    }

    @Override
    public void initWithNiwsConfig(IClientConfig clientConfig) {
    }
}

除此之外，IPing還有如下幾種實(shí)現(xiàn)：

image.png

3.3.3 Spring Cloud集成下的IPing實(shí)現(xiàn):NIWSDiscoveryPing

而和Spring Cloud 集成后，IPing的默認(rèn)實(shí)現(xiàn)，是NIWSDiscoveryPing ,其使用Eureka作為服務(wù)注冊(cè)和發(fā)現(xiàn)例获，則校驗(yàn)服務(wù)是否可用，則通過(guò)監(jiān)聽(tīng)Eureka 服務(wù)更新來(lái)更新Ribbon的Server狀態(tài)曹仗，而具體的實(shí)現(xiàn)就是 NIWSDiscoveryPing:

/**
 * "Ping" Discovery Client
 * i.e. we dont do a real "ping". We just assume that the server is up if Discovery Client says so
 * @author stonse
 *
 */
public class NIWSDiscoveryPing extends AbstractLoadBalancerPing {
            
        BaseLoadBalancer lb = null; 
        

        public NIWSDiscoveryPing() {
        }
        
        public BaseLoadBalancer getLb() {
            return lb;
        }

        /**
         * Non IPing interface method - only set this if you care about the "newServers Feature"
         * @param lb
         */
        public void setLb(BaseLoadBalancer lb) {
            this.lb = lb;
        }

        public boolean isAlive(Server server) {
            boolean isAlive = true;
         //取 Eureka Server 的Instance實(shí)例狀態(tài)作為Ribbon服務(wù)的狀態(tài)   
                 if (server!=null && server instanceof DiscoveryEnabledServer){
                DiscoveryEnabledServer dServer = (DiscoveryEnabledServer)server;                
                InstanceInfo instanceInfo = dServer.getInstanceInfo();
                if (instanceInfo!=null){                    
                    InstanceStatus status = instanceInfo.getStatus();
                    if (status!=null){
                        isAlive = status.equals(InstanceStatus.UP);
                    }
                }
            }
            return isAlive;
        }

        @Override
        public void initWithNiwsConfig(
                IClientConfig clientConfig) {
        }
        
}

Spring Cloud下的默認(rèn)實(shí)現(xiàn)入口：

    @Bean
    @ConditionalOnMissingBean
    public IPing ribbonPing(IClientConfig config) {
        if (this.propertiesFactory.isSet(IPing.class, serviceId)) {
            return this.propertiesFactory.get(IPing.class, config, serviceId);
        }
        NIWSDiscoveryPing ping = new NIWSDiscoveryPing();
        ping.initWithNiwsConfig(config);
        return ping;
    }

3.4 如何從服務(wù)列表中挑選一個(gè)合適的服務(wù)實(shí)例榨汤？

3.4.1 服務(wù)實(shí)例容器：ServerList的維護(hù)

負(fù)載均衡器通過(guò) ServerList來(lái)統(tǒng)一維護(hù)服務(wù)實(shí)例，具體模式如下：

image.png

基礎(chǔ)的接口定義非常簡(jiǎn)單：

/**
 * Interface that defines the methods sed to obtain the List of Servers
 * @author stonse
 *
 * @param <T>
 */
public interface ServerList<T extends Server> {
    //獲取初始化的服務(wù)列表
    public List<T> getInitialListOfServers();
    
    /**
     * Return updated list of servers. This is called say every 30 secs
     * (configurable) by the Loadbalancer's Ping cycle
     * 獲取更新后的的服務(wù)列表
     */
    public List<T> getUpdatedListOfServers();   

}

在Ribbon的實(shí)現(xiàn)中整葡，在ServerList中件余，維護(hù)著Server的實(shí)例，并返回最新的List<Server>集合，供LoadBalancer使用

image.png

ServerList<Server>的職能：
負(fù)責(zé)維護(hù)服務(wù)實(shí)例啼器，并使用ServerListFilter過(guò)濾器過(guò)濾出符合要求的服務(wù)實(shí)例列表List<Server>

3.4.2 服務(wù)實(shí)例列表過(guò)濾器ServerListFilter

服務(wù)實(shí)例列表過(guò)濾器ServerListFilter的職能很簡(jiǎn)單：
傳入一個(gè)服務(wù)實(shí)例列表旬渠，過(guò)濾出滿(mǎn)足過(guò)濾條件的服務(wù)列表

public interface ServerListFilter<T extends Server> {
    public List<T> getFilteredListOfServers(List<T> servers);
}

3.4.2.1 Ribbon 的默認(rèn)ServerListFilter實(shí)現(xiàn)：ZoneAffinityServerListFilter

Ribbon默認(rèn)采取了區(qū)域優(yōu)先的過(guò)濾策略，即當(dāng)Server列表中端壳，過(guò)濾出和當(dāng)前實(shí)例所在的區(qū)域（zone）一致的server列表
與此相關(guān)聯(lián)的告丢，Ribbon有兩個(gè)相關(guān)得配置參數(shù)：

其核心實(shí)現(xiàn)如下所示：

public class ZoneAffinityServerListFilter<T extends Server> extends
        AbstractServerListFilter<T> implements IClientConfigAware {

     @Override
    public List<T> getFilteredListOfServers(List<T> servers) {
          //zone非空损谦，并且開(kāi)啟了區(qū)域優(yōu)先岖免，并且服務(wù)實(shí)例數(shù)量不為空
          if (zone != null && (zoneAffinity || zoneExclusive) && servers !=null && servers.size() > 0){
           //基于斷言過(guò)濾服務(wù)列表 
           List<T> filteredServers = Lists.newArrayList(Iterables.filter(
                    servers, this.zoneAffinityPredicate.getServerOnlyPredicate()));
            //如果允許區(qū)域優(yōu)先，則返回過(guò)濾列表
            if (shouldEnableZoneAffinity(filteredServers)) {
                return filteredServers;
            } else if (zoneAffinity) {
                overrideCounter.increment();
            }
        }
        return servers;
    }
    // 判斷是否應(yīng)該使用區(qū)域優(yōu)先過(guò)濾條件
    private boolean shouldEnableZoneAffinity(List<T> filtered) {    
        if (!zoneAffinity && !zoneExclusive) {
            return false;
        }
        if (zoneExclusive) {
            return true;
        }
       // 獲取統(tǒng)計(jì)信息
        LoadBalancerStats stats = getLoadBalancerStats();
        if (stats == null) {
            return zoneAffinity;
        } else {
            logger.debug("Determining if zone affinity should be enabled with given server list: {}", filtered);
            //獲取區(qū)域Server快照照捡，包含統(tǒng)計(jì)數(shù)據(jù)
            ZoneSnapshot snapshot = stats.getZoneSnapshot(filtered);
            //平均負(fù)載颅湘，此負(fù)載的意思是，當(dāng)前所有的Server中栗精，平均每臺(tái)機(jī)器上的活躍請(qǐng)求數(shù)
           double loadPerServer = snapshot.getLoadPerServer();
            int instanceCount = snapshot.getInstanceCount();            
            int circuitBreakerTrippedCount = snapshot.getCircuitTrippedCount();
            // 1. 如果Server斷路的比例超過(guò)了設(shè)置的上限(默認(rèn)`0.8`)
            // 2. 或者當(dāng)前負(fù)載超過(guò)了設(shè)置的負(fù)載上限
            // 3. 如果可用的服務(wù)小于設(shè)置的服務(wù)上限`默認(rèn)為2`
            if (((double) circuitBreakerTrippedCount) / instanceCount >= blackOutServerPercentageThreshold.get() 
                    || loadPerServer >= activeReqeustsPerServerThreshold.get()
                    || (instanceCount - circuitBreakerTrippedCount) < availableServersThreshold.get()) {
                logger.debug("zoneAffinity is overriden. blackOutServerPercentage: {}, activeReqeustsPerServer: {}, availableServers: {}", 
                        new Object[] {(double) circuitBreakerTrippedCount / instanceCount,  loadPerServer, instanceCount - circuitBreakerTrippedCount});
                return false;
            } else {
                return true;
            }
            
        }
    }
        
   

}

具體判斷流程如下所示：

image.png

3.4.2.2 Ribbon 的ServerListFilter實(shí)現(xiàn)2：ZonePreferenceServerListFilter

ZonePreferenceServerListFilter 集成自 ZoneAffinityServerListFilter闯参，在此基礎(chǔ)上做了拓展，在 ZoneAffinityServerListFilter返回結(jié)果的基礎(chǔ)上悲立，再過(guò)濾出和本地服務(wù)相同區(qū)域(zone)的服務(wù)列表鹿寨。

核心邏輯-什么時(shí)候起作用？
當(dāng)指定了當(dāng)前服務(wù)的所在Zone薪夕，并且 ZoneAffinityServerListFilter 沒(méi)有起到過(guò)濾效果時(shí)脚草，ZonePreferenceServerListFilter會(huì)返回當(dāng)前Zone的Server列表。

public class ZonePreferenceServerListFilter extends ZoneAffinityServerListFilter<Server> {

    private String zone;

    @Override
    public void initWithNiwsConfig(IClientConfig niwsClientConfig) {
        super.initWithNiwsConfig(niwsClientConfig);
        if (ConfigurationManager.getDeploymentContext() != null) {
            this.zone = ConfigurationManager.getDeploymentContext().getValue(
                    ContextKey.zone);
        }
    }

    @Override
    public List<Server> getFilteredListOfServers(List<Server> servers) {
        //父類(lèi)的基礎(chǔ)上原献，獲取過(guò)濾結(jié)果
                List<Server> output = super.getFilteredListOfServers(servers);
        //沒(méi)有起到過(guò)濾效果馏慨，則進(jìn)行區(qū)域優(yōu)先過(guò)濾
                if (this.zone != null && output.size() == servers.size()) {
            List<Server> local = new ArrayList<>();
            for (Server server : output) {
                if (this.zone.equalsIgnoreCase(server.getZone())) {
                    local.add(server);
                }
            }
            if (!local.isEmpty()) {
                return local;
            }
        }
        return output;
    }

    public String getZone() {
        return zone;
    }

    public void setZone(String zone) {
        this.zone = zone;
    }

3.4.2.3 Ribbon 的ServerListFilter實(shí)現(xiàn)3：ServerListSubsetFilter

這個(gè)過(guò)濾器作用于當(dāng)Server數(shù)量列表特別龐大時(shí)（比如有上百個(gè)Server實(shí)例），這時(shí)嚼贡，長(zhǎng)時(shí)間保持Http鏈接也不太合適熏纯，可以適當(dāng)?shù)乇Ａ舨糠址?wù)，舍棄其中一些服務(wù)粤策，這樣可使釋放沒(méi)必要的鏈接。
此過(guò)濾器也是繼承自 ZoneAffinityServerListFilter,在此基礎(chǔ)上做了拓展误窖，在實(shí)際使用中不太常見(jiàn)叮盘，這個(gè)后續(xù)再展開(kāi)介紹，暫且不表霹俺。

3.4.3 LoadBalancer選擇服務(wù)實(shí)例 的流程

LoadBalancer的核心功能是根據(jù)負(fù)載情況柔吼，從服務(wù)列表中挑選最合適的服務(wù)實(shí)例。LoadBalancer內(nèi)部采用了如下圖所示的組件完成：

image.png

LoadBalancer 選擇服務(wù)實(shí)例的流程

1. 通過(guò)ServerList獲取當(dāng)前可用的服務(wù)實(shí)例列表丙唧；
2. 通過(guò)ServerListFilter將步驟1 得到的服務(wù)列表進(jìn)行一次過(guò)濾愈魏，返回滿(mǎn)足過(guò)濾器條件的服務(wù)實(shí)例列表；
3. 應(yīng)用Rule規(guī)則，結(jié)合服務(wù)實(shí)例的統(tǒng)計(jì)信息,返回滿(mǎn)足規(guī)則的某一個(gè)服務(wù)實(shí)例培漏；

通過(guò)上述的流程可以看到溪厘，實(shí)際上，在服務(wù)實(shí)例列表選擇的過(guò)程中牌柄，有兩次過(guò)濾的機(jī)會(huì)：第一次是首先通過(guò)ServerListFilter過(guò)濾器畸悬，另外一次是用過(guò)IRule 的選擇規(guī)則進(jìn)行過(guò)濾

通過(guò)ServerListFilter進(jìn)行服務(wù)實(shí)例過(guò)濾的策略上面已經(jīng)介紹得比較詳細(xì)了,接下來(lái)將介紹Rule是如何從一堆服務(wù)列表中選擇服務(wù)的。
在介紹Rule之前珊佣，需要介紹一個(gè)概念：Server統(tǒng)計(jì)信息

當(dāng)LoadBalancer在選擇合適的Server提供給應(yīng)用后蹋宦，應(yīng)用會(huì)向該Server發(fā)送服務(wù)請(qǐng)求，則在請(qǐng)求的過(guò)程中咒锻，應(yīng)用會(huì)根據(jù)請(qǐng)求的相應(yīng)時(shí)間或者網(wǎng)絡(luò)連接情況等進(jìn)行統(tǒng)計(jì)冷冗；當(dāng)應(yīng)用后續(xù)從LoadBalancer選擇合適的Server時(shí)，LoadBalancer 會(huì)根據(jù)每個(gè)服務(wù)的統(tǒng)計(jì)信息惑艇，結(jié)合Rule來(lái)判定哪個(gè)服務(wù)是最合適的贾惦。

3.4.3.1 負(fù)載均衡器LoaderBalancer 都統(tǒng)計(jì)了哪些關(guān)于服務(wù)實(shí)例Server相關(guān)的信息？

3.4.3.2 服務(wù)斷路器的工作原理

當(dāng)有某個(gè)服務(wù)存在多個(gè)實(shí)例時(shí)，在請(qǐng)求的過(guò)程中秽荤，負(fù)載均衡器會(huì)統(tǒng)計(jì)每次請(qǐng)求的情況（請(qǐng)求相應(yīng)時(shí)間甜奄，是否發(fā)生網(wǎng)絡(luò)異常等），當(dāng)出現(xiàn)了請(qǐng)求出現(xiàn)累計(jì)重試時(shí)窃款，負(fù)載均衡器會(huì)標(biāo)識(shí)當(dāng)前服務(wù)實(shí)例课兄，設(shè)置當(dāng)前服務(wù)實(shí)例的斷路的時(shí)間區(qū)間，在此區(qū)間內(nèi)晨继，當(dāng)請(qǐng)求過(guò)來(lái)時(shí)烟阐，負(fù)載均衡器會(huì)將此服務(wù)實(shí)例從可用服務(wù)實(shí)例列表中暫時(shí)剔除，優(yōu)先選擇其他服務(wù)實(shí)例紊扬。

相關(guān)統(tǒng)計(jì)信息如下：

@Monitor(name="CircuitBreakerTripped", type = DataSourceType.INFORMATIONAL)    
    public boolean isCircuitBreakerTripped() {
        return isCircuitBreakerTripped(System.currentTimeMillis());
    }
    
    public boolean isCircuitBreakerTripped(long currentTime) {
        //斷路器熔斷的時(shí)間戳
        long circuitBreakerTimeout = getCircuitBreakerTimeout();
        if (circuitBreakerTimeout <= 0) {
            return false;
        }
        return circuitBreakerTimeout > currentTime;//還在熔斷區(qū)間內(nèi)檀葛，則返回熔斷結(jié)果
    }
   //獲取熔斷超時(shí)時(shí)間
   private long getCircuitBreakerTimeout() {
        long blackOutPeriod = getCircuitBreakerBlackoutPeriod();
        if (blackOutPeriod <= 0) {
            return 0;
        }
        return lastConnectionFailedTimestamp + blackOutPeriod;
    }
    //返回中斷毫秒數(shù)
    private long getCircuitBreakerBlackoutPeriod() {
        int failureCount = successiveConnectionFailureCount.get();
        int threshold = connectionFailureThreshold.get();
       // 連續(xù)失敗，但是尚未超過(guò)上限腹缩，則服務(wù)中斷周期為 0 屿聋，表示可用 
       if (failureCount < threshold) {
            return 0;
        }
        //當(dāng)鏈接失敗超過(guò)閾值時(shí)空扎，將進(jìn)行熔斷，熔斷的時(shí)間間隔為：
        int diff = (failureCount - threshold) > 16 ? 16 : (failureCount - threshold);
        int blackOutSeconds = (1 << diff) * circuitTrippedTimeoutFactor.get();
        if (blackOutSeconds > maxCircuitTrippedTimeout.get()) {
            blackOutSeconds = maxCircuitTrippedTimeout.get();
        }
        return blackOutSeconds * 1000L;
    }

熔斷時(shí)間的計(jì)算

1. 計(jì)算累計(jì)連接失敗計(jì)數(shù)successiveConnectionFailureCount 是否超過(guò) 鏈接失敗閾值connectionFailureThreshold润讥。如果 successiveConnectionFailureCount < connectionFailureThreshold,即尚未超過(guò)限額转锈，則熔斷時(shí)間為 0 ；反之象对，如果超過(guò)限額黑忱，則進(jìn)行步驟2的計(jì)算；
2. 計(jì)算失敗基數(shù)勒魔，最大不得超過(guò) 16：diff = (failureCount - threshold) > 16 ? 16 : (failureCount - threshold)
3. 根據(jù)超時(shí)因子circuitTrippedTimeoutFactor計(jì)算超時(shí)時(shí)間： int blackOutSeconds = (1 << diff) * circuitTrippedTimeoutFactor.get();
4. 超時(shí)時(shí)間不得超過(guò)最大超時(shí)時(shí)間`maxCircuitTrippedTimeout 上線甫煞，

當(dāng)有鏈接失敗情況出現(xiàn)斷路邏輯時(shí)，將會(huì)最多：1<<16 * 10 =320s冠绢、最少1<<1 * 10 =100s 的請(qǐng)求熔斷時(shí)間抚吠，再此期間內(nèi)，此Server將會(huì)被忽略弟胀。
即：
熔斷時(shí)間最大值：1<<16 * 10 =320s
熔斷時(shí)間最小值：1<<1 * 10 =100s

熔斷統(tǒng)計(jì)何時(shí)清空楷力？

熔斷的統(tǒng)計(jì)有自己的清除策略，當(dāng)如下幾種場(chǎng)景存在時(shí)孵户，熔斷統(tǒng)計(jì)會(huì)清空歸零：

1. 當(dāng)請(qǐng)求時(shí)萧朝，發(fā)生的異常不是斷路攔截類(lèi)的異常（Exception）時(shí)(至于如何節(jié)點(diǎn)是否是斷路攔截類(lèi)異常，可以自定義)
2. 當(dāng)請(qǐng)求未發(fā)生異常夏哭，切且有結(jié)果返回時(shí)

3.4.3.3 定義IRule检柬，從服務(wù)實(shí)例列表中，選擇最合適的Server實(shí)例

image.png

由上圖可見(jiàn)竖配，IRule會(huì)從服務(wù)列表中何址，根據(jù)自身定義的規(guī)則，返回最合適的Server實(shí)例进胯，其接口定義如下：

public interface IRule{
    /*
     * choose one alive server from lb.allServers or
     * lb.upServers according to key
     * 
     * @return choosen Server object. NULL is returned if none
     *  server is available 
     */

    public Server choose(Object key);
    
    public void setLoadBalancer(ILoadBalancer lb);
    
    public ILoadBalancer getLoadBalancer();    
}

Ribbon定義了一些常見(jiàn)的規(guī)則

image.png

3.4.3.3.1 RoundRobinRule 的實(shí)現(xiàn)

public class RoundRobinRule extends AbstractLoadBalancerRule {

    private AtomicInteger nextServerCyclicCounter;
    private static final boolean AVAILABLE_ONLY_SERVERS = true;
    private static final boolean ALL_SERVERS = false;

    private static Logger log = LoggerFactory.getLogger(RoundRobinRule.class);

    public RoundRobinRule() {
        nextServerCyclicCounter = new AtomicInteger(0);
    }

    public RoundRobinRule(ILoadBalancer lb) {
        this();
        setLoadBalancer(lb);
    }

    public Server choose(ILoadBalancer lb, Object key) {
        if (lb == null) {
            log.warn("no load balancer");
            return null;
        }

        Server server = null;
        int count = 0;
        //10次重試機(jī)制
        while (server == null && count++ < 10) {
            List<Server> reachableServers = lb.getReachableServers();
            List<Server> allServers = lb.getAllServers();
            int upCount = reachableServers.size();
            int serverCount = allServers.size();

            if ((upCount == 0) || (serverCount == 0)) {
                log.warn("No up servers available from load balancer: " + lb);
                return null;
            }
            // 生成輪詢(xún)數(shù)據(jù)
            int nextServerIndex = incrementAndGetModulo(serverCount);
            server = allServers.get(nextServerIndex);

            if (server == null) {
                /* Transient. */
                Thread.yield();
                continue;
            }

            if (server.isAlive() && (server.isReadyToServe())) {
                return (server);
            }

            // Next.
            server = null;
        }

        if (count >= 10) {
            log.warn("No available alive servers after 10 tries from load balancer: "
                    + lb);
        }
        return server;
    }

    /**
     * Inspired by the implementation of {@link AtomicInteger#incrementAndGet()}.
     *
     * @param modulo The modulo to bound the value of the counter.
     * @return The next value.
     */
    private int incrementAndGetModulo(int modulo) {
        for (;;) {
            int current = nextServerCyclicCounter.get();
            int next = (current + 1) % modulo;
            if (nextServerCyclicCounter.compareAndSet(current, next))
                return next;
        }
    }

    @Override
    public Server choose(Object key) {
        return choose(getLoadBalancer(), key);
    }
}

3.4.3.3.2 ZoneAvoidanceRule的實(shí)現(xiàn)

ZoneAvoidanceRule的處理思路：

1. ZoneAvoidancePredicate 計(jì)算出哪個(gè)Zone的服務(wù)最差磁餐，然后將此Zone的服務(wù)從服務(wù)列表中剔除掉违崇；
2. AvailabilityPredicate 將處于熔斷狀態(tài)的服務(wù)剔除掉阿弃；
3. 將上述兩步驟過(guò)濾后的服務(wù)通過(guò)RoundRobinRule挑選一個(gè)服務(wù)實(shí)例返回

ZoneAvoidancePredicate 剔除最差的Zone的過(guò)程：

public static Set<String> getAvailableZones(
            Map<String, ZoneSnapshot> snapshot, double triggeringLoad,
            double triggeringBlackoutPercentage) {
        if (snapshot.isEmpty()) {
            return null;
        }
        Set<String> availableZones = new HashSet<String>(snapshot.keySet());
        if (availableZones.size() == 1) {
            return availableZones;
        }
        Set<String> worstZones = new HashSet<String>();
        double maxLoadPerServer = 0;
        boolean limitedZoneAvailability = false;

        for (Map.Entry<String, ZoneSnapshot> zoneEntry : snapshot.entrySet()) {
            String zone = zoneEntry.getKey();
            ZoneSnapshot zoneSnapshot = zoneEntry.getValue();
            int instanceCount = zoneSnapshot.getInstanceCount();
            if (instanceCount == 0) {
                availableZones.remove(zone);
                limitedZoneAvailability = true;
            } else {
                double loadPerServer = zoneSnapshot.getLoadPerServer();
                   //如果負(fù)載超過(guò)限額，則將用可用區(qū)中剔除出去
                  if (((double) zoneSnapshot.getCircuitTrippedCount())
                        / instanceCount >= triggeringBlackoutPercentage
                        || loadPerServer < 0) {
                    availableZones.remove(zone);
                    limitedZoneAvailability = true;
                } else {
                    //計(jì)算最差的Zone區(qū)域
                    if (Math.abs(loadPerServer - maxLoadPerServer) < 0.000001d) {
                        // they are the same considering double calculation
                        // round error
                        worstZones.add(zone);
                    } else if (loadPerServer > maxLoadPerServer) {
                        maxLoadPerServer = loadPerServer;
                        worstZones.clear();
                        worstZones.add(zone);
                    }
                }
            }
        }
        // 如果最大負(fù)載沒(méi)有超過(guò)上限羞延，則返回所有可用分區(qū)
        if (maxLoadPerServer < triggeringLoad && !limitedZoneAvailability) {
            // zone override is not needed here
            return availableZones;
        }
        // 從最差的可用分區(qū)中隨機(jī)挑選一個(gè)剔除渣淳，這么做是保證服務(wù)的高可用
        String zoneToAvoid = randomChooseZone(snapshot, worstZones);
        if (zoneToAvoid != null) {
            availableZones.remove(zoneToAvoid);
        }
        return availableZones;

    }

四. Ribbon的配置參數(shù)

五. 結(jié)語(yǔ)

Ribbon是Spring Cloud框架中相當(dāng)核心的模塊入愧，負(fù)責(zé)著服務(wù)負(fù)載調(diào)用，Ribbon也可以脫離SpringCloud單獨(dú)使用嗤谚。
另外Ribbon是客戶(hù)端的負(fù)載均衡框架棺蛛，即每個(gè)客戶(hù)端上，獨(dú)立維護(hù)著自身的調(diào)用信息統(tǒng)計(jì)巩步，相互隔離旁赊；也就是說(shuō)：Ribbon的負(fù)載均衡表現(xiàn)在各個(gè)機(jī)器上變現(xiàn)并不完全一致
Ribbon 也是整個(gè)組件框架中最復(fù)雜的一環(huán)，控制流程上為了保證服務(wù)的高可用性椅野，有很多比較細(xì)節(jié)的參數(shù)控制终畅，在使用的過(guò)程中，需要深入理清每個(gè)環(huán)節(jié)的處理機(jī)制竟闪，這樣在問(wèn)題定位上會(huì)高效很多离福。

亦山札記，聚焦微服務(wù)炼蛤、中間件