增加自定义全链路随机权重

Alleninggx · Dec 29, 2019 · 09b55fb · 09b55fb
1 parent 2dbdcfb
commit 09b55fb
Show file tree

Hide file tree

Showing 4 changed files with 53 additions and 93 deletions.
diff --git a/...n/discovery/plugin/framework/loadbalance/weight/AbstractArrayWeightRandomLoadBalance.java b/...n/discovery/plugin/framework/loadbalance/weight/AbstractArrayWeightRandomLoadBalance.java
@@ -10,14 +10,15 @@
  */
 
 import java.util.List;
-import java.util.concurrent.ThreadLocalRandom;
 
 import org.apache.commons.collections4.CollectionUtils;
 
 import com.nepxion.discovery.plugin.framework.loadbalance.WeightRandomLoadBalance;
 import com.netflix.loadbalancer.Server;
 
 public abstract class AbstractArrayWeightRandomLoadBalance<T> implements WeightRandomLoadBalance<T> {
+    private ArrayWeightRandom arrayWeightRandom = new ArrayWeightRandom();
+
     @Override
     public Server choose(List<Server> serverList, T t) {
         if (CollectionUtils.isEmpty(serverList)) {
@@ -30,36 +31,8 @@ public Server choose(List<Server> serverList, T t) {
             weights[i] = getWeight(server, t);
         }
 
-        int index = getIndex(weights);
+        int index = arrayWeightRandom.getIndex(weights);
 
         return serverList.get(index);
     }
-
-    private int getIndex(int[] weights) {
-        // 次序号/权重区间值
-        int[][] weightHolder = new int[weights.length][2];
-        // 总权重
-        int totalWeight = 0;
-        // 赋值次序号和区间值累加的数组值，从小到大排列
-        // 例如，对于权重分别为20，40， 60的三个服务，将形成[0, 20)，[20, 60)，[60, 120]三个区间
-        for (int i = 0; i < weights.length; i++) {
-            if (weights[i] <= 0) {
-                continue;
-            }
-
-            totalWeight += weights[i];
-            weightHolder[i][0] = i;
-            weightHolder[i][1] = totalWeight;
-        }
-
-        // 获取介于0(含)和n(不含)伪随机，均匀分布的int值
-        int hitWeight = ThreadLocalRandom.current().nextInt(totalWeight) + 1; // [1, totalWeight)
-        for (int i = 0; i < weightHolder.length; i++) {
-            if (hitWeight <= weightHolder[i][1]) {
-                return weightHolder[i][0];
-            }
-        }
-
-        return weightHolder[0][0];
-    }
 }
diff --git a/...ain/java/com/nepxion/discovery/plugin/framework/loadbalance/weight/ArrayWeightRandom.java b/...ain/java/com/nepxion/discovery/plugin/framework/loadbalance/weight/ArrayWeightRandom.java
@@ -0,0 +1,42 @@
+package com.nepxion.discovery.plugin.framework.loadbalance.weight;
+
+/**
+ * <p>Title: Nepxion Discovery</p>
+ * <p>Description: Nepxion Discovery</p>
+ * <p>Copyright: Copyright (c) 2017-2050</p>
+ * <p>Company: Nepxion</p>
+ * @author Haojun Ren
+ * @version 1.0
+ */
+
+import java.util.concurrent.ThreadLocalRandom;
+
+public class ArrayWeightRandom {
+    public int getIndex(int[] weights) {
+        // 次序号/权重区间值
+        int[][] weightHolder = new int[weights.length][2];
+        // 总权重
+        int totalWeight = 0;
+        // 赋值次序号和区间值累加的数组值，从小到大排列
+        // 例如，对于权重分别为20，40， 60的三个服务，将形成[0, 20)，[20, 60)，[60, 120]三个区间
+        for (int i = 0; i < weights.length; i++) {
+            if (weights[i] <= 0) {
+                continue;
+            }
+
+            totalWeight += weights[i];
+            weightHolder[i][0] = i;
+            weightHolder[i][1] = totalWeight;
+        }
+
+        // 获取介于0(含)和n(不含)伪随机，均匀分布的int值
+        int hitWeight = ThreadLocalRandom.current().nextInt(totalWeight) + 1; // [1, totalWeight)
+        for (int i = 0; i < weightHolder.length; i++) {
+            if (hitWeight <= weightHolder[i][1]) {
+                return weightHolder[i][0];
+            }
+        }
+
+        return weightHolder[0][0];
+    }
+}
diff --git a/.../com/nepxion/discovery/plugin/framework/loadbalance/ArrayWeightRandomLoadBalanceTest.java b/.../com/nepxion/discovery/plugin/framework/loadbalance/ArrayWeightRandomLoadBalanceTest.java
@@ -14,10 +14,11 @@
 import java.util.HashMap;
 import java.util.List;
 import java.util.Map;
-import java.util.concurrent.ThreadLocalRandom;
 
 import org.apache.commons.collections4.CollectionUtils;
 
+import com.nepxion.discovery.plugin.framework.loadbalance.weight.ArrayWeightRandom;
+
 public class ArrayWeightRandomLoadBalanceTest {
     public static void main(String[] args) {
         test(1000000);
@@ -51,7 +52,7 @@ public static void test(int totalCount) {
         int v4Count = 0;
         int v5Count = 0;
         for (int i = 0; i < totalCount; i++) {
-            String server = choose(serverList, weightMap);
+            String server = random(serverList, weightMap);
             if (server.startsWith("1.0")) {
                 v1Count++;
             }
@@ -81,7 +82,9 @@ public static void test(int totalCount) {
         System.out.println("------------------------------");
     }
 
-    public static String choose(List<String> serverList, Map<String, Integer> weightMap) {
+    private static ArrayWeightRandom arrayWeightRandom = new ArrayWeightRandom();
+
+    public static String random(List<String> serverList, Map<String, Integer> weightMap) {
         if (CollectionUtils.isEmpty(serverList)) {
             return null;
         }
@@ -92,36 +95,8 @@ public static String choose(List<String> serverList, Map<String, Integer> weight
             weights[i] = weightMap.get(server);
         }
 
-        int index = getIndex(weights);
+        int index = arrayWeightRandom.getIndex(weights);
 
         return serverList.get(index);
     }
-
-    private static int getIndex(int[] weights) {
-        // 次序号/权重区间值
-        int[][] weightHolder = new int[weights.length][2];
-        // 总权重
-        int totalWeight = 0;
-        // 赋值次序号和区间值累加的数组值，从小到大排列
-        // 例如，对于权重分别为20，40， 60的三个服务，将形成[0, 20)，[20, 60)，[60, 120]三个区间
-        for (int i = 0; i < weights.length; i++) {
-            if (weights[i] <= 0) {
-                continue;
-            }
-
-            totalWeight += weights[i];
-            weightHolder[i][0] = i;
-            weightHolder[i][1] = totalWeight;
-        }
-
-        // 获取介于0(含)和n(不含)伪随机，均匀分布的int值
-        int hitWeight = ThreadLocalRandom.current().nextInt(totalWeight) + 1; // [1, totalWeight)
-        for (int i = 0; i < weightHolder.length; i++) {
-            if (hitWeight <= weightHolder[i][1]) {
-                return weightHolder[i][0];
-            }
-        }
-
-        return weightHolder[0][0];
-    }
 }
diff --git a/...va/com/nepxion/discovery/plugin/framework/loadbalance/MapWeightRandomLoadBalanceTest.java b/...va/com/nepxion/discovery/plugin/framework/loadbalance/MapWeightRandomLoadBalanceTest.java
@@ -12,14 +12,11 @@
 import java.text.DecimalFormat;
 import java.util.ArrayList;
 import java.util.List;
-import java.util.SortedMap;
-import java.util.TreeMap;
 
-import org.apache.commons.collections4.MapUtils;
 import org.apache.commons.lang3.tuple.ImmutablePair;
 import org.apache.commons.lang3.tuple.Pair;
 
-import com.nepxion.discovery.common.exception.DiscoveryException;
+import com.nepxion.discovery.plugin.framework.loadbalance.weight.MapWeightRandom;
 
 public class MapWeightRandomLoadBalanceTest {
     public static void main(String[] args) {
@@ -77,31 +74,4 @@ public static void test(int totalCount) {
         System.out.println("耗时时间：" + (System.currentTimeMillis() - t));
         System.out.println("------------------------------");
     }
-
-    public static class MapWeightRandom<K, V extends Number> {
-        private TreeMap<Double, K> weightMap = new TreeMap<Double, K>();
-
-        public MapWeightRandom(List<Pair<K, V>> pairlist) {
-            for (Pair<K, V> pair : pairlist) {
-                double value = pair.getValue().doubleValue();
-                if (value <= 0) {
-                    continue;
-                }
-
-                double lastWeight = weightMap.size() == 0 ? 0 : weightMap.lastKey().doubleValue();
-                weightMap.put(value + lastWeight, pair.getKey());
-            }
-        }
-
-        public K random() {
-            if (MapUtils.isEmpty(weightMap)) {
-                throw new DiscoveryException("Weight values are all <= 0 or invalid format");
-            }
-
-            double randomWeight = weightMap.lastKey() * Math.random();
-            SortedMap<Double, K> tailMap = weightMap.tailMap(randomWeight, false);
-
-            return weightMap.get(tailMap.firstKey());
-        }
-    }
 }