释放双眼，带上耳机，听听看~！
jedis是一个著名的key-value存储系统，而作为其官方推荐的java版客户端jedis也非常强大和稳定，支持事务、管道及有jedis自身实现的分布式。
在这里对jedis关于事务、管道和分布式的调用方式做一个简单的介绍和对比：
一、普通同步方式
最简单和基础的调用方式，
2
3
4
5
6
7
8
9
10
11
12
2public void test1Normal() {
3 Jedis jedis = new Jedis("localhost");
4 long start = System.currentTimeMillis();
5 for (int i = 0; i < 100000; i++) {
6 String result = jedis.set("n" + i, "n" + i);
7 }
8 long end = System.currentTimeMillis();
9 System.out.println("Simple SET: " + ((end - start)/1000.0) + " seconds");
10 jedis.disconnect();
11}
12
很简单吧，每次set之后都可以返回结果，标记是否成功。
二、事务方式(Transactions)
redis的事务很简单，他主要目的是保障，一个client发起的事务中的命令可以连续的执行，而中间不会插入其他client的命令。
看下面例子：
2
3
4
5
6
7
8
9
10
11
12
13
14
2public void test2Trans() {
3 Jedis jedis = new Jedis("localhost");
4 long start = System.currentTimeMillis();
5 Transaction tx = jedis.multi();
6 for (int i = 0; i < 100000; i++) {
7 tx.set("t" + i, "t" + i);
8 }
9 List<Object> results = tx.exec();
10 long end = System.currentTimeMillis();
11 System.out.println("Transaction SET: " + ((end - start)/1000.0) + " seconds");
12 jedis.disconnect();
13}
14
我们调用jedis.watch(…)方法来监控key，如果调用后key值发生变化，则整个事务会执行失败。另外，事务中某个操作失败，并不会回滚其他操作。这一点需要注意。还有，我们可以使用discard()方法来取消事务。
三、管道(Pipelining)
有时，我们需要采用异步方式，一次发送多个指令，不同步等待其返回结果。这样可以取得非常好的执行效率。这就是管道，调用方法如下：
2
3
4
5
6
7
8
9
10
11
12
13
14
2public void test3Pipelined() {
3 Jedis jedis = new Jedis("localhost");
4 Pipeline pipeline = jedis.pipelined();
5 long start = System.currentTimeMillis();
6 for (int i = 0; i < 100000; i++) {
7 pipeline.set("p" + i, "p" + i);
8 }
9 List<Object> results = pipeline.syncAndReturnAll();
10 long end = System.currentTimeMillis();
11 System.out.println("Pipelined SET: " + ((end - start)/1000.0) + " seconds");
12 jedis.disconnect();
13}
14
四、管道中调用事务
就Jedis提供的方法而言，是可以做到在管道中使用事务，其代码如下：
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
2public void test4combPipelineTrans() {
3 jedis = new Jedis("localhost");
4 long start = System.currentTimeMillis();
5 Pipeline pipeline = jedis.pipelined();
6 pipeline.multi();
7 for (int i = 0; i < 100000; i++) {
8 pipeline.set("" + i, "" + i);
9 }
10 pipeline.exec();
11 List<Object> results = pipeline.syncAndReturnAll();
12 long end = System.currentTimeMillis();
13 System.out.println("Pipelined transaction: " + ((end - start)/1000.0) + " seconds");
14 jedis.disconnect();
15}
16
但是经测试（见本文后续部分），发现其效率和单独使用事务差不多，甚至还略微差点。
五、分布式直连同步调用
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
2public void test5shardNormal() {
3 List<JedisShardInfo> shards = Arrays.asList(
4 new JedisShardInfo("localhost",6379),
5 new JedisShardInfo("localhost",6380));
6
7 ShardedJedis sharding = new ShardedJedis(shards);
8
9 long start = System.currentTimeMillis();
10 for (int i = 0; i < 100000; i++) {
11 String result = sharding.set("sn" + i, "n" + i);
12 }
13 long end = System.currentTimeMillis();
14 System.out.println("Simple@Sharing SET: " + ((end - start)/1000.0) + " seconds");
15
16 sharding.disconnect();
17}
18
这个是分布式直接连接，并且是同步调用，每步执行都返回执行结果。类似地，还有异步管道调用。
六、分布式直连异步调用
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
2public void test6shardpipelined() {
3 List<JedisShardInfo> shards = Arrays.asList(
4 new JedisShardInfo("localhost",6379),
5 new JedisShardInfo("localhost",6380));
6
7 ShardedJedis sharding = new ShardedJedis(shards);
8
9 ShardedJedisPipeline pipeline = sharding.pipelined();
10 long start = System.currentTimeMillis();
11 for (int i = 0; i < 100000; i++) {
12 pipeline.set("sp" + i, "p" + i);
13 }
14 List<Object> results = pipeline.syncAndReturnAll();
15 long end = System.currentTimeMillis();
16 System.out.println("Pipelined@Sharing SET: " + ((end - start)/1000.0) + " seconds");
17
18 sharding.disconnect();
19}
20
七、分布式连接池同步调用
如果，你的分布式调用代码是运行在线程中，那么上面两个直连调用方式就不合适了，因为直连方式是非线程安全的，这个时候，你就必须选择连接池调用。
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
2public void test7shardSimplePool() {
3 List<JedisShardInfo> shards = Arrays.asList(
4 new JedisShardInfo("localhost",6379),
5 new JedisShardInfo("localhost",6380));
6
7 ShardedJedisPool pool = new ShardedJedisPool(new JedisPoolConfig(), shards);
8
9 ShardedJedis one = pool.getResource();
10
11 long start = System.currentTimeMillis();
12 for (int i = 0; i < 100000; i++) {
13 String result = one.set("spn" + i, "n" + i);
14 }
15 long end = System.currentTimeMillis();
16 pool.returnResource(one);
17 System.out.println("Simple@Pool SET: " + ((end - start)/1000.0) + " seconds");
18
19 pool.destroy();
20}
21
上面是同步方式，当然还有异步方式。
八、分布式连接池异步调用
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
2public void test8shardPipelinedPool() {
3 List<JedisShardInfo> shards = Arrays.asList(
4 new JedisShardInfo("localhost",6379),
5 new JedisShardInfo("localhost",6380));
6
7 ShardedJedisPool pool = new ShardedJedisPool(new JedisPoolConfig(), shards);
8
9 ShardedJedis one = pool.getResource();
10
11 ShardedJedisPipeline pipeline = one.pipelined();
12
13 long start = System.currentTimeMillis();
14 for (int i = 0; i < 100000; i++) {
15 pipeline.set("sppn" + i, "n" + i);
16 }
17 List<Object> results = pipeline.syncAndReturnAll();
18 long end = System.currentTimeMillis();
19 pool.returnResource(one);
20 System.out.println("Pipelined@Pool SET: " + ((end - start)/1000.0) + " seconds");
21 pool.destroy();
22}
23
九、需要注意的地方
事务和管道都是异步模式。在事务和管道中不能同步查询结果。比如下面两个调用，都是不允许的：
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
2 for (int i = 0; i < 100000; i++) {
3 tx.set("t" + i, "t" + i);
4 }
5 System.out.println(tx.get("t1000").get()); //不允许
6
7 List<Object> results = tx.exec();
8
9 …
10 …
11
12 Pipeline pipeline = jedis.pipelined();
13 long start = System.currentTimeMillis();
14 for (int i = 0; i < 100000; i++) {
15 pipeline.set("p" + i, "p" + i);
16 }
17 System.out.println(pipeline.get("p1000").get()); //不允许
18
19 List<Object> results = pipeline.syncAndReturnAll();
20
事务和管道都是异步的，个人感觉，在管道中再进行事务调用，没有必要，不如直接进行事务模式。
分布式中，连接池的性能比直连的性能略好(见后续测试部分)。
分布式调用中不支持事务。
因为事务是在服务器端实现，而在分布式中，每批次的调用对象都可能访问不同的机器，所以，没法进行事务。
十、测试
运行上面的代码，进行测试，其结果如下：
2
3
4
5
6
7
8
9
10
11
12
2
3Transaction SET: 0.5 seconds
4Pipelined SET: 0.353 seconds
5Pipelined transaction: 0.509 seconds
6
7Simple@Sharing SET: 5.289 seconds
8Pipelined@Sharing SET: 0.348 seconds
9
10Simple@Pool SET: 5.039 seconds
11Pipelined@Pool SET: 0.401 seconds
12
另外，经测试分布式中用到的机器越多，调用会越慢。上面是2片，下面是5片：
2
3
4
5
2Pipelined@Sharing SET: 0.51 seconds
3Simple@Pool SET: 5.223 seconds
4Pipelined@Pool SET: 0.518 seconds
5
下面是10片：
2
3
4
5
2Pipelined@Sharing SET: 0.794 seconds
3Simple@Pool SET: 5.624 seconds
4Pipelined@Pool SET: 0.762 seconds
5
下面是100片：
2
3
4
5
2Pipelined@Sharing SET: 8.185 seconds
3Simple@Pool SET: 13.29 seconds
4Pipelined@Pool SET: 7.767 seconds
5
分布式中，连接池方式调用不但线程安全外，根据上面的测试数据，也可以看出连接池比直连的效率更好。
十一、完整的测试代码
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
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
2
3import java.util.Arrays;
4import java.util.List;
5
6import org.junit.AfterClass;
7import org.junit.BeforeClass;
8import org.junit.Test;
9
10import redis.clients.jedis.Jedis;
11import redis.clients.jedis.JedisPoolConfig;
12import redis.clients.jedis.JedisShardInfo;
13import redis.clients.jedis.Pipeline;
14import redis.clients.jedis.ShardedJedis;
15import redis.clients.jedis.ShardedJedisPipeline;
16import redis.clients.jedis.ShardedJedisPool;
17import redis.clients.jedis.Transaction;
18
19import org.junit.FixMethodOrder;
20import org.junit.runners.MethodSorters;
21
22@FixMethodOrder(MethodSorters.NAME_ASCENDING)
23public class TestJedis {
24
25 private static Jedis jedis;
26 private static ShardedJedis sharding;
27 private static ShardedJedisPool pool;
28
29 @BeforeClass
30 public static void setUpBeforeClass() throws Exception {
31 List<JedisShardInfo> shards = Arrays.asList(
32 new JedisShardInfo("localhost",6379),
33 new JedisShardInfo("localhost",6379)); //使用相同的ip:port,仅作测试
34
35
36 jedis = new Jedis("localhost");
37 sharding = new ShardedJedis(shards);
38
39 pool = new ShardedJedisPool(new JedisPoolConfig(), shards);
40 }
41
42 @AfterClass
43 public static void tearDownAfterClass() throws Exception {
44 jedis.disconnect();
45 sharding.disconnect();
46 pool.destroy();
47 }
48
49 @Test
50 public void test1Normal() {
51 long start = System.currentTimeMillis();
52 for (int i = 0; i < 100000; i++) {
53 String result = jedis.set("n" + i, "n" + i);
54 }
55 long end = System.currentTimeMillis();
56 System.out.println("Simple SET: " + ((end - start)/1000.0) + " seconds");
57 }
58
59 @Test
60 public void test2Trans() {
61 long start = System.currentTimeMillis();
62 Transaction tx = jedis.multi();
63 for (int i = 0; i < 100000; i++) {
64 tx.set("t" + i, "t" + i);
65 }
66 //System.out.println(tx.get("t1000").get());
67
68 List<Object> results = tx.exec();
69 long end = System.currentTimeMillis();
70 System.out.println("Transaction SET: " + ((end - start)/1000.0) + " seconds");
71 }
72
73 @Test
74 public void test3Pipelined() {
75 Pipeline pipeline = jedis.pipelined();
76 long start = System.currentTimeMillis();
77 for (int i = 0; i < 100000; i++) {
78 pipeline.set("p" + i, "p" + i);
79 }
80 //System.out.println(pipeline.get("p1000").get());
81 List<Object> results = pipeline.syncAndReturnAll();
82 long end = System.currentTimeMillis();
83 System.out.println("Pipelined SET: " + ((end - start)/1000.0) + " seconds");
84 }
85
86 @Test
87 public void test4combPipelineTrans() {
88 long start = System.currentTimeMillis();
89 Pipeline pipeline = jedis.pipelined();
90 pipeline.multi();
91 for (int i = 0; i < 100000; i++) {
92 pipeline.set("" + i, "" + i);
93 }
94 pipeline.exec();
95 List<Object> results = pipeline.syncAndReturnAll();
96 long end = System.currentTimeMillis();
97 System.out.println("Pipelined transaction: " + ((end - start)/1000.0) + " seconds");
98 }
99
100 @Test
101 public void test5shardNormal() {
102 long start = System.currentTimeMillis();
103 for (int i = 0; i < 100000; i++) {
104 String result = sharding.set("sn" + i, "n" + i);
105 }
106 long end = System.currentTimeMillis();
107 System.out.println("Simple@Sharing SET: " + ((end - start)/1000.0) + " seconds");
108 }
109
110 @Test
111 public void test6shardpipelined() {
112 ShardedJedisPipeline pipeline = sharding.pipelined();
113 long start = System.currentTimeMillis();
114 for (int i = 0; i < 100000; i++) {
115 pipeline.set("sp" + i, "p" + i);
116 }
117 List<Object> results = pipeline.syncAndReturnAll();
118 long end = System.currentTimeMillis();
119 System.out.println("Pipelined@Sharing SET: " + ((end - start)/1000.0) + " seconds");
120 }
121
122 @Test
123 public void test7shardSimplePool() {
124 ShardedJedis one = pool.getResource();
125
126 long start = System.currentTimeMillis();
127 for (int i = 0; i < 100000; i++) {
128 String result = one.set("spn" + i, "n" + i);
129 }
130 long end = System.currentTimeMillis();
131 pool.returnResource(one);
132 System.out.println("Simple@Pool SET: " + ((end - start)/1000.0) + " seconds");
133 }
134
135 @Test
136 public void test8shardPipelinedPool() {
137 ShardedJedis one = pool.getResource();
138
139 ShardedJedisPipeline pipeline = one.pipelined();
140
141 long start = System.currentTimeMillis();
142 for (int i = 0; i < 100000; i++) {
143 pipeline.set("sppn" + i, "n" + i);
144 }
145 List<Object> results = pipeline.syncAndReturnAll();
146 long end = System.currentTimeMillis();
147 pool.returnResource(one);
148 System.out.println("Pipelined@Pool SET: " + ((end - start)/1000.0) + " seconds");
149 }
150}
151
