memcached演练(6) 高可用实例HA(伪集群方案 )

本系列文章<memcached的演练>,这是第6篇,前面文章,已经阐述了,memcached的安装,访问,session管理,存储管理。从本篇开始,就分开有几篇演练下memcached的高可用相关。

memcached的安全,承载着后台数据库的巨大访问,意义重大。

简单介绍下HA的梗概,如果时间允许尽量都演练下个方案,然后横向比较下各方案的优缺点。

memcached演练(6) 高可用实例HA(伪集群方案 )

本篇主要内容

伪集群方案的测试

明确memcached的高可用方案

memcached演练(6) 高可用实例HA(伪集群方案 )

通过比较memcached和redis两种NOSQL的方案,很容易发现,memcached的集群方案设计很简单,主要在client实现。

伪集群方案测试步骤

  1. 准备演练环境

主机
memcached主目录
端口
192.168.163.146 (hadoop1)
/usr/local/memcached/
11211
192.168.163.156 (hadoop2) /usr/local/memcached/ 11211
192.168.163.166 (hadoop3) /usr/local/memcached/ 11211

安装不久罗列了,请参照《memcached演练(1) 搭建memcached服务》

2.准备测试代码

1
2
3
4
5
6
7
8
9
10
11
12
13
    public void testSet() throws ExecutionException, InterruptedException, IOException {
        final MemcachedClient mcc =  new MemcachedClient(new BinaryConnectionFactory(),AddrUtil.getAddresses("192.168.163.146:11211 192.168.163.156:11211 192.168.163.166:11211"));
        for(int i=0;i<50;i++){
            mcc.set("clusterKey_"+StringUtils.leftPad(""+i,4,"0"), 19000"clusterValue_"+i);
        }
        System.out.println("set ok");
 
        for(int i=0;i<50;i++){
            Object o = mcc.get("clusterKey_" + StringUtils.leftPad(""+i,4,"0"));
            System.out.println(o);
        }
//        mcc.shutdown();
    }

3.测试结果分析

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
[[email protected] scripts]# ./memcached-tool hadoop1:11211 dump |sort
Dumping memcache contents
  Number of buckets: 1
  Number of items  : 16
Dumping bucket 1 - 16 total items
add clusterKey_0002 0 1471374793 14
add clusterKey_0005 0 1471374793 14
add clusterKey_0008 0 1471374793 14
add clusterKey_0011 0 1471374793 15
add clusterKey_0014 0 1471374793 15
add clusterKey_0017 0 1471374793 15
add clusterKey_0020 0 1471374793 15
add clusterKey_0023 0 1471374793 15
add clusterKey_0026 0 1471374793 15
add clusterKey_0029 0 1471374793 15
add clusterKey_0032 0 1471374793 15
add clusterKey_0035 0 1471374793 15
add clusterKey_0038 0 1471374793 15
add clusterKey_0041 0 1471374793 15
add clusterKey_0044 0 1471374793 15
add clusterKey_0047 0 1471374793 15
 
[[email protected] scripts]# ./memcached-tool hadoop2:11211 dump |sort
Dumping memcache contents
  Number of buckets: 1
  Number of items  : 17
Dumping bucket 1 - 17 total items
add clusterKey_0000 0 1471374795 14
add clusterKey_0003 0 1471374795 14
add clusterKey_0006 0 1471374795 14
add clusterKey_0009 0 1471374795 14
add clusterKey_0012 0 1471374795 15
add clusterKey_0015 0 1471374795 15
add clusterKey_0018 0 1471374795 15
add clusterKey_0021 0 1471374795 15
add clusterKey_0024 0 1471374795 15
add clusterKey_0027 0 1471374795 15
add clusterKey_0030 0 1471374795 15
add clusterKey_0033 0 1471374795 15
add clusterKey_0036 0 1471374795 15
add clusterKey_0039 0 1471374795 15
add clusterKey_0042 0 1471374795 15
add clusterKey_0045 0 1471374795 15
add clusterKey_0048 0 1471374795 15
 
[[email protected] scripts]# ./memcached-tool hadoop3:11211 dump |sort
Dumping memcache contents
  Number of buckets: 1
  Number of items  : 17
Dumping bucket 1 - 17 total items
add clusterKey_0001 0 1471374794 14
add clusterKey_0004 0 1471374794 14
add clusterKey_0007 0 1471374794 14
add clusterKey_0010 0 1471374794 15
add clusterKey_0013 0 1471374794 15
add clusterKey_0016 0 1471374794 15
add clusterKey_0019 0 1471374794 15
add clusterKey_0022 0 1471374794 15
add clusterKey_0025 0 1471374794 15
add clusterKey_0028 0 1471374794 15
add clusterKey_0031 0 1471374794 15
add clusterKey_0034 0 1471374794 15
add clusterKey_0037 0 1471374794 15
add clusterKey_0040 0 1471374794 15
add clusterKey_0043 0 1471374794 15
add clusterKey_0046 0 1471374794 15
add clusterKey_0049 0 1471374794 15

上面数据,简单整理成图,很直观,数据分布很分散的。这样的好处,可以减少热点。

memcached演练(6) 高可用实例HA(伪集群方案 )


值得注意的是

1
 final MemcachedClient mcc =  new MemcachedClient(new BinaryConnectionFactory(),AddrUtil.getAddresses("192.168.163.146:11211 192.168.163.156:11211 192.168.163.166:11211"));

AddrUtil.getAddresses参数,可以重复添加节点信息,这就相当于机器有权重的感觉了,明显会造成数据分布不均匀。

4.测试下权重情况下的数据分布

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
public class SpyMemcachedClusterTest extends TestCase
{
 
    public void testSet() throws ExecutionException, InterruptedException, IOException {
        final MemcachedClient mcc =  new MemcachedClient(new BinaryConnectionFactory(),AddrUtil.getAddresses("192.168.163.146:11211 192.168.163.146:11211 192.168.163.146:11211 192.168.163.156:11211 192.168.163.156:11211 192.168.163.166:11211"));
        for(int i=0;i<150;i++){
            mcc.set("clusterKey_"+StringUtils.leftPad(""+i,4,"0"), 19000"clusterValue_"+i);
        }
        System.out.println("ok");
//        mcc.shutdown();
        for(int i=0;i<150;i++){
            Object o = mcc.get("clusterKey_" + StringUtils.leftPad(""+i,4,"0"));
            System.out.println(o);
        }
 
    }
}

测试结果截图

memcached演练(6) 高可用实例HA(伪集群方案 )

两图比较:是不是很容易,就看出来了。一致性可以从数据分布方面考虑,还是非常平均的,但数据量的分布,就和权重有关系。基本就是 3:2:1。和上面程序代码中配置的权重基本一致。

继续做实验。

5.剔出一个节点,看看命中率

1
2
3
4
5
6
7
8
9
10
11
12
13
public void testGets() throws ExecutionException, InterruptedException, IOException{
    final MemcachedClient mcc =  new MemcachedClient(new BinaryConnectionFactory(),AddrUtil.getAddresses(" 192.168.163.146:11211 192.168.163.146:11211 192.168.163.156:11211 192.168.163.156:11211 192.168.163.166:11211"));
    int nohit=0;
    for(int i=0;i<150;i++){
        String key = "clusterKey_" + StringUtils.leftPad("" + i, 4"0");
        Object o = mcc.get(key);
        if(o==null){
            System.out.println(key+"未命中");
            nohit++;
        }
    }
    System.out.println("丢失:"+nohit);
}

修改内容,hadoop1:hadoop2:hadoop3,有原来的 3:2:1分布,修改成2:2:1分布。

测试结果

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
clusterKey_0001未命中    
clusterKey_0002未命中    
clusterKey_0010未命中    
clusterKey_0011未命中    
clusterKey_0019未命中    
clusterKey_0020未命中    
clusterKey_0028未命中    
clusterKey_0037未命中    
clusterKey_0039未命中    
clusterKey_0046未命中    
clusterKey_0048未命中    
clusterKey_0049未命中    
clusterKey_0055未命中    
clusterKey_0057未命中    
clusterKey_0058未命中    
clusterKey_0064未命中    
clusterKey_0066未命中    
clusterKey_0067未命中    
clusterKey_0069未命中    
clusterKey_0073未命中    
clusterKey_0075未命中    
clusterKey_0076未命中    
clusterKey_0078未命中    
clusterKey_0079未命中    
clusterKey_0082未命中    
clusterKey_0084未命中    
clusterKey_0085未命中    
clusterKey_0087未命中    
clusterKey_0088未命中    
clusterKey_0089未命中    
clusterKey_0091未命中    
clusterKey_0093未命中    
clusterKey_0094未命中    
clusterKey_0096未命中    
clusterKey_0097未命中    
clusterKey_0098未命中    
clusterKey_0099未命中    
clusterKey_0100未命中    
clusterKey_0101未命中    
clusterKey_0109未命中    
clusterKey_0110未命中    
clusterKey_0118未命中    
clusterKey_0127未命中    
clusterKey_0129未命中    
clusterKey_0136未命中    
clusterKey_0138未命中    
clusterKey_0139未命中    
clusterKey_0145未命中    
clusterKey_0147未命中    
clusterKey_0148未命中    
丢失:50

丢失了1/3。

即使改成 3:2,丢失也基本上到达1/3。 这个问题需要重视,否则非常容易产生缓存穿透情况。

在以后,会研究下spymemcached的一致性算法。



本文转自 randy_shandong 51CTO博客,原文链接:http://blog.51cto.com/dba10g/1839379,如需转载请自行联系原作者