架构师训练营第五周作业

用户头像
张明森
关注
发布于: 2020 年 07 月 05 日

实现一致性 hash 算法



哈希算法:

public interface HashStrategy {
long getHashCode(String key);
}



import java.nio.ByteBuffer;
import java.nio.ByteOrder;
public class MurmurHashStrategy implements HashStrategy {
@Override
public long getHashCode(String key) {
ByteBuffer buf = ByteBuffer.wrap(key.getBytes());
int seed = 0x1234ABCD;
ByteOrder byteOrder = buf.order();
buf.order(ByteOrder.LITTLE_ENDIAN);
long m = 0xc6a4a7935bd1e995L;
int r = 47;
long h = seed ^ (buf.remaining() * m);
long k;
while (buf.remaining() >= 8) {
k = buf.getLong();
k *= m;
k ^= k >>> r;
k *= m;
h ^= k;
h *= m;
}
if (buf.remaining() > 0) {
ByteBuffer finish = ByteBuffer.allocate(8).order(
ByteOrder.LITTLE_ENDIAN);
finish.put(buf).rewind();
h ^= finish.getLong();
h *= m;
}
h ^= h >>> r;
h *= m;
h ^= h >>> r;
buf.order(byteOrder);
return (int) (h & 0xffffffffL);
}
}

一致性哈希:

import java.util.List;
import java.util.Map;
import java.util.TreeMap;
public class ConsistentHash {
private String name;
public String getName() {
return name;
}
/**
* 虚拟节点数量
*/
private int virtualNodeSize;
public int getVirtualNodeSize() {
return virtualNodeSize;
}
private String VIRTUAL_NODE_SUFFIX="_";
private TreeMap<Long,Node> virtualNodes=new TreeMap<>();
private HashStrategy hashStrategy;
public ConsistentHash(List<Node> nodes,HashStrategy hashStrategy,int virtualNodeSize) {
this.hashStrategy=hashStrategy;
name=hashStrategy.getClass().getSimpleName();
this.virtualNodeSize=virtualNodeSize;
buildVirtualNodes(nodes);
}
private void buildVirtualNodes(List<Node> nodes){
nodes.forEach(it->{
for(int i=0;i<virtualNodeSize;i++){
String key=it.toString()+VIRTUAL_NODE_SUFFIX+i;
long hash = hashStrategy.getHashCode(key);
virtualNodes.put(hash,it);
}
});
}
public Node getNode(String dataKey){
if(virtualNodes==null)return null;
long hash=hashStrategy.getHashCode(dataKey);
if(!virtualNodes.containsKey(hash)){
Map.Entry<Long,Node> entry=virtualNodes.ceilingEntry(hash);
if(entry!=null){
return entry.getValue();
}else{
return virtualNodes.firstEntry().getValue();
}
}else{
return virtualNodes.get(hash);
}
}
}



编写测试用例测试这个算法

测试 100 万 KV 数据,10 个服务器节点的情况下,计算这些 KV 数据在服务器上分布数量的标准差,以评估算法的存储负载不均衡性。

@RunWith(JUnit4.class)
public class TestHash {
@Test
public void testHash(){
int circleCount=1000;
int dataNum=1000000;
ExecutorService executorService = new ThreadPoolExecutor(5, 10,
0, TimeUnit.SECONDS,
new ArrayBlockingQueue<>(circleCount),
new ThreadPoolExecutor.DiscardPolicy());// 指定拒绝策略
Map<Integer,Integer> map=new Hashtable<>();
try {
CountDownLatch cc=new CountDownLatch(circleCount);
for (int w=0;w<circleCount;w++) {
executorService.execute(new Runnable() {
@Override
public void run() {
List<Node> nodeList=new ArrayList<>();
for(int i=0;i<10;i++){
Node node=new Node();
Random r=new Random();
node.setIp(r.nextInt(255)+"."+r.nextInt(255)+"."+r.nextInt(255)+"."+r.nextInt(255));
node.setPort("3306");
nodeList.add(node);
}
List<String> testDatas=new ArrayList<>(dataNum);
for(int j=0;j<dataNum;j++){
String uuid= UUID.randomUUID().toString();
testDatas.add(uuid);
}
int kk=0;
double min=1000000.0;
for(int k=150;k<=200;k=k+10){
ConsistentHash consistentHashk=new ConsistentHash(nodeList,new MurmurHashStrategy(),k);
double r=statistics(testDatas,consistentHashk);
if(r<min){
min=r;
kk=k;
}
}
System.out.println("虚拟节点数量:"+kk+",平均差:"+min);
map.put(kk,(map.get(kk)==null?0:map.get(kk))+1);
cc.countDown();
}
});
}
cc.await();
System.out.println("");
System.out.println("------------------------------");
map.entrySet().forEach(it->{
System.out.println("虚拟节点数量:"+it.getKey().toString() + ",平均差最小次数:" +it.getValue());
});
} catch (InterruptedException e) {
}
}
public double standardDeviation(Long[] x) {
int m = x.length;
double sum = 0;
for (int i = 0; i < m; i++) {//求和
sum += x[i];
}
double dAve = sum / m;//求平均值
double dVar = 0;
for (int i = 0; i < m; i++) {//求方差
dVar += (x[i] - dAve) * (x[i] - dAve);
}
return Math.sqrt(dVar / m);
}
private double statistics(List<String> testDatas, ConsistentHash consistentHash){
Map<Node,Long> dataMap=new HashMap<>();
for(int j=0;j<testDatas.size();j++){
String uuid= testDatas.get(j);
Node node=consistentHash.getNode(uuid);
dataMap.put(node,dataMap.get(node)==null?0:dataMap.get(node)+1);
}
Collection<Long> datas=dataMap.values();
double result=standardDeviation(datas.toArray(new Long[0]));
return result;
}
}

测试结果

------------------------------
虚拟节点数量:150,平均差最小次数:126
虚拟节点数量:160,平均差最小次数:125
虚拟节点数量:170,平均差最小次数:110
虚拟节点数量:180,平均差最小次数:117
虚拟节点数量:190,平均差最小次数:171
虚拟节点数量:200,平均差最小次数:351

循环1000次,虚拟节点是200的时候,平均差最小出现的次数最多



用户头像

张明森

关注

还未添加个人签名 2017.10.16 加入

还未添加个人简介

评论

发布
暂无评论
架构师训练营第五周作业