ZYallers / consistent_hash

一致性哈希算法实现数据分布, 它支持动态添加、删除节点, 并引入虚拟节点保证数据在节点间的均匀分布

consistent_hash

一致性哈希算法实现数据分布, 它支持动态添加、删除节点, 并引入虚拟节点保证数据在节点间的均匀分布。

特点

1. 动态添加、删除节点
2. 引入虚拟节点保证数据均匀分布
3. 自定义hash函数

扩容数据分布

func TestConsistentHash_DataDistribution(t *testing.T) {
	ch := NewConsistentHash(300, nil)
	ch.AddNode("db1", "db2", "db3", "db4")

	dataTotal := 50000
	nodesHash := make(map[string]int)
	for i := 0; i < dataTotal; i++ {
		dataKey := "data" + strconv.Itoa(i)
		node := ch.GetNode(dataKey)
		nodesHash[node]++
	}
	for k, v := range nodesHash {
		t.Logf("%s count: %d (%.2f%%)\n", k, v, float64(v)/float64(dataTotal)*100)
	}

	t.Log("--------------扩容后---------------")
	ch.AddNode("db5", "db6")

	countPrint := func(total, incr int) int {
		t.Logf("--------------数据增长%d---------------\n", incr)
		dataTotal := incr + total
		for i := total; i < dataTotal; i++ {
			dataKey := "data" + strconv.Itoa(i)
			node := ch.GetNode(dataKey)
			nodesHash[node]++
		}
		for k, v := range nodesHash {
			t.Logf("%s count: %d (%.2f%%)\n", k, v, float64(v)/float64(dataTotal)*100)
		}
		return dataTotal
	}

	dataTotal = countPrint(dataTotal, 100000)
	dataTotal = countPrint(dataTotal, 300000)
	dataTotal = countPrint(dataTotal, 500000)
}

测试结果：

db1 count: 12220 (24.44%)
db3 count: 13211 (26.42%)
db4 count: 12153 (24.31%)
db2 count: 12416 (24.83%)
--------------扩容后---------------
--------------数据增长100000---------------
db6 count: 15861 (10.57%)
db5 count: 16160 (10.77%)
db1 count: 29222 (19.48%)
db3 count: 31524 (21.02%)
db4 count: 29421 (19.61%)
db2 count: 27812 (18.54%)
--------------数据增长300000---------------
db6 count: 63630 (14.14%)
db5 count: 64501 (14.33%)
db1 count: 80014 (17.78%)
db3 count: 85481 (19.00%)
db4 count: 81780 (18.17%)
db2 count: 74594 (16.58%)
--------------数据增长500000---------------
db5 count: 145391 (15.30%)
db1 count: 164249 (17.29%)
db3 count: 175432 (18.47%)
db4 count: 169568 (17.85%)
db2 count: 152161 (16.02%)
db6 count: 143199 (15.07%)

License

Released under the MIT License