dubbo负载均衡策略及对应源码分析

在集群负载均衡时，Dubbo 提供了多种均衡策略，缺省为 random 随机调用。我们还可以扩展自己的负责均衡策略，前提是你已经从一个小白变成了大牛，嘻嘻

1、Random LoadBalance

         1.1  随机，按权重设置随机概率。

         1.2  在一个截面上碰撞的概率高，但调用量越大分布越均匀，而且按概率使用权重后也比较均匀，有利于动态调整提供者权重。

         1.3 源码分析

           

package com.alibaba.dubbo.rpc.cluster.loadbalance;

import java.util.List;
import java.util.Random;

import com.alibaba.dubbo.common.URL;
import com.alibaba.dubbo.rpc.Invocation;
import com.alibaba.dubbo.rpc.Invoker;

/**
 * random load balance.
 *
 * @author qianlei
 * @author william.liangf
 */
public class RandomLoadBalance extends AbstractLoadBalance {

    public static final String NAME = "random";

    private final Random random = new Random();

    protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
        int length = invokers.size(); // 总个数
        int totalWeight = 0; // 总权重
        boolean sameWeight = true; // 权重是否都一样
        for (int i = 0; i < length; i++) {
            int weight = getWeight(invokers.get(i), invocation);
            totalWeight += weight; // 累计总权重
            if (sameWeight && i > 0
                    && weight != getWeight(invokers.get(i - 1), invocation)) {
                sameWeight = false; // 计算所有权重是否一样
            }
        }
        if (totalWeight > 0 && ! sameWeight) {
            // 如果权重不相同且权重大于0则按总权重数随机
            int offset = random.nextInt(totalWeight);
            // 并确定随机值落在哪个片断上
            for (int i = 0; i < length; i++) {
                offset -= getWeight(invokers.get(i), invocation);
                if (offset < 0) {
                    return invokers.get(i);
                }
            }
        }
        // 如果权重相同或权重为0则均等随机
        return invokers.get(random.nextInt(length));
    }

}

       说明：从源码可以看出随机负载均衡的策略分为两种情况

         a. 如果总权重大于0并且权重不相同，就生成一个1~totalWeight(总权重数)的随机数，然后再把随机数和所有的权重值一一相减得到一个新的随机数，直到随机 数小于0，那么此时访问的服务器就是使得随机数小于0的权重所在的机器

         b.  如果权重相同或者总权重数为0，就生成一个1~length(权重的总个数)的随机数，此时所访问的机器就是这个随机数对应的权重所在的机器

2、RoundRobin LoadBalance

      2.1 轮循，按公约后的权重设置轮循比率。

      2.2 存在慢的提供者累积请求的问题，比如：第二台机器很慢，但没挂，当请求调到第二台时就卡在那，久而久之，所有请求都卡在调到第二台上。

      2.3 源码分析

      

package com.alibaba.dubbo.rpc.cluster.loadbalance;

import java.util.ArrayList;
import java.util.List;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ConcurrentMap;

import com.alibaba.dubbo.common.URL;
import com.alibaba.dubbo.common.utils.AtomicPositiveInteger;
import com.alibaba.dubbo.rpc.Invocation;
import com.alibaba.dubbo.rpc.Invoker;

/**
 * Round robin load balance.
 *
 * @author qian.lei
 * @author william.liangf
 */
public class RoundRobinLoadBalance extends AbstractLoadBalance {

    public static final String NAME = "roundrobin"; 
    
    private final ConcurrentMap<String, AtomicPositiveInteger> sequences = new ConcurrentHashMap<String, AtomicPositiveInteger>();

    private final ConcurrentMap<String, AtomicPositiveInteger> weightSequences = new ConcurrentHashMap<String, AtomicPositiveInteger>();

    protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
        String key = invokers.get(0).getUrl().getServiceKey() + "." + invocation.getMethodName();
        int length = invokers.size(); // 总个数
        int maxWeight = 0; // 最大权重
        int minWeight = Integer.MAX_VALUE; // 最小权重
        for (int i = 0; i < length; i++) {
            int weight = getWeight(invokers.get(i), invocation);
            maxWeight = Math.max(maxWeight, weight); // 累计最大权重
            minWeight = Math.min(minWeight, weight); // 累计最小权重
        }
        if (maxWeight > 0 && minWeight < maxWeight) { // 权重不一样
            AtomicPositiveInteger weightSequence = weightSequences.get(key);
            if (weightSequence == null) {
                weightSequences.putIfAbsent(key, new AtomicPositiveInteger());
                weightSequence = weightSequences.get(key);
            }
            int currentWeight = weightSequence.getAndIncrement() % maxWeight;
            List<Invoker<T>> weightInvokers = new ArrayList<Invoker<T>>();
            for (Invoker<T> invoker : invokers) { // 筛选权重大于当前权重基数的Invoker
                if (getWeight(invoker, invocation) > currentWeight) {
                    weightInvokers.add(invoker);
                }
            }
            int weightLength = weightInvokers.size();
            if (weightLength == 1) {
                return weightInvokers.get(0);
            } else if (weightLength > 1) {
                invokers = weightInvokers;
                length = invokers.size();
            }
        }
        AtomicPositiveInteger sequence = sequences.get(key);
        if (sequence == null) {
            sequences.putIfAbsent(key, new AtomicPositiveInteger());
            sequence = sequences.get(key);
        }
        // 取模轮循
        return invokers.get(sequence.getAndIncrement() % length);
    }

}

      说明：从源码可以看出轮循负载均衡的算法是：

                 a.  如果权重不一样时，获取一个当前的权重基数，然后从权重集合中筛选权重大于当前权重基数的集合，如果筛选出的集合的长度为1，此时所访问的机器就是集合里面的权重对应的机器

                 b.  如果权重一样时就取模轮循

3、LeastActive LoadBalance

        3.1 最少活跃调用数，相同活跃数的随机，活跃数指调用前后计数差(调用前的时刻减去响应后的时刻的值)。

        3.2 使慢的提供者收到更少请求，因为越慢的提供者的调用前后计数差会越大

        3.3 对应的源码

       

package com.alibaba.dubbo.rpc.cluster.loadbalance;

import java.util.List;
import java.util.Random;

import com.alibaba.dubbo.common.Constants;
import com.alibaba.dubbo.common.URL;
import com.alibaba.dubbo.rpc.Invocation;
import com.alibaba.dubbo.rpc.Invoker;
import com.alibaba.dubbo.rpc.RpcStatus;

/**
 * LeastActiveLoadBalance
 * 
 * @author william.liangf
 */
public class LeastActiveLoadBalance extends AbstractLoadBalance {

    public static final String NAME = "leastactive";
    
    private final Random random = new Random();

    protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
        int length = invokers.size(); // 总个数
        int leastActive = -1; // 最小的活跃数
        int leastCount = 0; // 相同最小活跃数的个数
        int[] leastIndexs = new int[length]; // 相同最小活跃数的下标
        int totalWeight = 0; // 总权重
        int firstWeight = 0; // 第一个权重，用于于计算是否相同
        boolean sameWeight = true; // 是否所有权重相同
        for (int i = 0; i < length; i++) {
            Invoker<T> invoker = invokers.get(i);
            int active = RpcStatus.getStatus(invoker.getUrl(), invocation.getMethodName()).getActive(); // 活跃数
            int weight = invoker.getUrl().getMethodParameter(invocation.getMethodName(), Constants.WEIGHT_KEY, Constants.DEFAULT_WEIGHT); // 权重
            if (leastActive == -1 || active < leastActive) { // 发现更小的活跃数，重新开始
                leastActive = active; // 记录最小活跃数
                leastCount = 1; // 重新统计相同最小活跃数的个数
                leastIndexs[0] = i; // 重新记录最小活跃数下标
                totalWeight = weight; // 重新累计总权重
                firstWeight = weight; // 记录第一个权重
                sameWeight = true; // 还原权重相同标识
            } else if (active == leastActive) { // 累计相同最小的活跃数
                leastIndexs[leastCount ++] = i; // 累计相同最小活跃数下标
                totalWeight += weight; // 累计总权重
                // 判断所有权重是否一样
                if (sameWeight && i > 0 
                        && weight != firstWeight) {
                    sameWeight = false;
                }
            }
        }
        // assert(leastCount > 0)
        if (leastCount == 1) {
            // 如果只有一个最小则直接返回
            return invokers.get(leastIndexs[0]);
        }
        if (! sameWeight && totalWeight > 0) {
            // 如果权重不相同且权重大于0则按总权重数随机
            int offsetWeight = random.nextInt(totalWeight);
            // 并确定随机值落在哪个片断上
            for (int i = 0; i < leastCount; i++) {
                int leastIndex = leastIndexs[i];
                offsetWeight -= getWeight(invokers.get(leastIndex), invocation);
                if (offsetWeight <= 0)
                    return invokers.get(leastIndex);
            }
        }
        // 如果权重相同或权重为0则均等随机
        return invokers.get(leastIndexs[random.nextInt(leastCount)]);
    }
}

    说明：源码里面的注释已经很清晰了，大致的意思就是活跃数越小的的机器分配到的请求越多

 4、ConsistentHash LoadBalance

       4.1 一致性 Hash，相同参数的请求总是发到同一提供者。

       4.2 当某一台提供者挂时，原本发往该提供者的请求，基于虚拟节点，平摊到其它提供者，不会引起剧烈变动。

       4.3 缺省只对第一个参数 Hash，如果要修改，请配置 <dubbo:parameter key="hash.arguments" value="0,1" />

       4.4 缺省用 160 份虚拟节点，如果要修改，请配置 <dubbo:parameter key="hash.nodes" value="320" />

       4.5 源码分析

/*
 * Copyright 1999-2012 Alibaba Group.
 *  
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *  
 *      http://www.apache.org/licenses/LICENSE-2.0
 *  
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
package com.alibaba.dubbo.rpc.cluster.loadbalance;

import java.io.UnsupportedEncodingException;
import java.security.MessageDigest;
import java.security.NoSuchAlgorithmException;
import java.util.List;
import java.util.SortedMap;
import java.util.TreeMap;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ConcurrentMap;

import com.alibaba.dubbo.common.Constants;
import com.alibaba.dubbo.common.URL;
import com.alibaba.dubbo.rpc.Invocation;
import com.alibaba.dubbo.rpc.Invoker;

/**
 * ConsistentHashLoadBalance
 * 
 * @author william.liangf
 */
public class ConsistentHashLoadBalance extends AbstractLoadBalance {

    private final ConcurrentMap<String, ConsistentHashSelector<?>> selectors = new ConcurrentHashMap<String, ConsistentHashSelector<?>>();

    @SuppressWarnings("unchecked")
    @Override
    protected <T> Invoker<T> doSelect(List<Invoker<T>> invokers, URL url, Invocation invocation) {
        String key = invokers.get(0).getUrl().getServiceKey() + "." + invocation.getMethodName();
        int identityHashCode = System.identityHashCode(invokers);
        ConsistentHashSelector<T> selector = (ConsistentHashSelector<T>) selectors.get(key);
        if (selector == null || selector.getIdentityHashCode() != identityHashCode) {
            selectors.put(key, new ConsistentHashSelector<T>(invokers, invocation.getMethodName(), identityHashCode));
            selector = (ConsistentHashSelector<T>) selectors.get(key);
        }
        return selector.select(invocation);
    }

    private static final class ConsistentHashSelector<T> {

        private final TreeMap<Long, Invoker<T>> virtualInvokers;

        private final int                       replicaNumber;
        
        private final int                       identityHashCode;
        
        private final int[]                     argumentIndex;

        public ConsistentHashSelector(List<Invoker<T>> invokers, String methodName, int identityHashCode) {
            this.virtualInvokers = new TreeMap<Long, Invoker<T>>();
            this.identityHashCode = System.identityHashCode(invokers);
            URL url = invokers.get(0).getUrl();
            this.replicaNumber = url.getMethodParameter(methodName, "hash.nodes", 160);
            String[] index = Constants.COMMA_SPLIT_PATTERN.split(url.getMethodParameter(methodName, "hash.arguments", "0"));
            argumentIndex = new int[index.length];
            for (int i = 0; i < index.length; i ++) {
                argumentIndex[i] = Integer.parseInt(index[i]);
            }
            for (Invoker<T> invoker : invokers) {
                for (int i = 0; i < replicaNumber / 4; i++) {
                    byte[] digest = md5(invoker.getUrl().toFullString() + i);
                    for (int h = 0; h < 4; h++) {
                        long m = hash(digest, h);
                        virtualInvokers.put(m, invoker);
                    }
                }
            }
        }

        public int getIdentityHashCode() {
            return identityHashCode;
        }

        public Invoker<T> select(Invocation invocation) {
            String key = toKey(invocation.getArguments());
            byte[] digest = md5(key);
            Invoker<T> invoker = sekectForKey(hash(digest, 0));
            return invoker;
        }

        private String toKey(Object[] args) {
            StringBuilder buf = new StringBuilder();
            for (int i : argumentIndex) {
                if (i >= 0 && i < args.length) {
                    buf.append(args[i]);
                }
            }
            return buf.toString();
        }

        private Invoker<T> sekectForKey(long hash) {
            Invoker<T> invoker;
            Long key = hash;
            if (!virtualInvokers.containsKey(key)) {
                SortedMap<Long, Invoker<T>> tailMap = virtualInvokers.tailMap(key);
                if (tailMap.isEmpty()) {
                    key = virtualInvokers.firstKey();
                } else {
                    key = tailMap.firstKey();
                }
            }
            invoker = virtualInvokers.get(key);
            return invoker;
        }

        private long hash(byte[] digest, int number) {
            return (((long) (digest[3 + number * 4] & 0xFF) << 24)
                    | ((long) (digest[2 + number * 4] & 0xFF) << 16)
                    | ((long) (digest[1 + number * 4] & 0xFF) << 8) 
                    | (digest[0 + number * 4] & 0xFF)) 
                    & 0xFFFFFFFFL;
        }

        private byte[] md5(String value) {
            MessageDigest md5;
            try {
                md5 = MessageDigest.getInstance("MD5");
            } catch (NoSuchAlgorithmException e) {
                throw new IllegalStateException(e.getMessage(), e);
            }
            md5.reset();
            byte[] bytes = null;
            try {
                bytes = value.getBytes("UTF-8");
            } catch (UnsupportedEncodingException e) {
                throw new IllegalStateException(e.getMessage(), e);
            }
            md5.update(bytes);
            return md5.digest();
        }

    }

}

说明：根据传递的参数进行hash然后调用服务，如果两次传递的参数一样就调用的是同一个机器上的服务

5、dubbo官方的文档的负载均衡配置示例

    服务端服务级别

   <dubbo:service interface="..." loadbalance="roundrobin" />

    客户端服务级别

   <dubbo:reference interface="..." loadbalance="roundrobin" />

    服务端方法级别

  <dubbo:service interface="...">
      <dubbo:method name="..." loadbalance="roundrobin"/>
  </dubbo:service>

    客户端方法级别

  <dubbo:reference interface="...">
      <dubbo:method name="..." loadbalance="roundrobin"/>
  </dubbo:reference>