kylin分析引擎：运行原理、维度和Cube等名词解释、核心算法

快来打我* 2023-10-05 13:08 25阅读 0赞

## 架构 ##

kylin是一款分析引擎，最常用的数据源是Hive。1.5版本之后数据源增加了kafka，但最常用的仍然是Hive。但是Hive查询比较慢，Kylin解决了这一问题。

**大致流程：**

第一步：从Hive或Kafka中获取数据;

第二步：有Cube构建引擎对数据进行加工;

第三步：将加工后的数据存储到HBase;（Kylin快的根本所在）

第四步：暴露接口供外部查询。

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70][]

以下这个组件负责数据的查询，亿级数据亚秒级别查询就靠它了：

![20201103165133576.png][]

以下两个组件用于解决离线数据的存储：

![20201103165043561.png][]

Kylin从Hive中获取到数据，然后由下面这个组件进行加工：将所有维度可能的组合给提前预计算好之后存储到HBase，查询九直接走HBase。这个预计算的过程就是MR来完成的，1.6之后可以选择spark作为计算引擎。

![2020110316500528.png][]

## 名词解释 ##

### 维度和度量 ###

比如：性别、年龄是维度，人数就是度量（也叫指标等）

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 1][]

### Cube和Cuboid ###

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 2][]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 3][]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 4][]

## 核心算法 ##

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 5][]

### 逐层构建算法 ###

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 6][]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 7][]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 8][]

![20201103212248800.png][]

### 快速构建算法 ###

![20201103212531911.png][]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 9][]

![20201103213309826.png][]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/2020110316564575.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[20201103165133576.png]: https://img-blog.csdnimg.cn/20201103165133576.png
[20201103165043561.png]: https://img-blog.csdnimg.cn/20201103165043561.png
[2020110316500528.png]: https://img-blog.csdnimg.cn/2020110316500528.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 1]: https://img-blog.csdnimg.cn/20201103170228593.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 2]: https://img-blog.csdnimg.cn/20201103170334746.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 3]: https://img-blog.csdnimg.cn/2020110317052749.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 4]: https://img-blog.csdnimg.cn/20201103170815286.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 5]: https://img-blog.csdnimg.cn/20201103171504333.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 6]: https://img-blog.csdnimg.cn/20201103211942316.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 7]: https://img-blog.csdnimg.cn/20201103212122184.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 8]: https://img-blog.csdnimg.cn/20201103212206577.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[20201103212248800.png]: https://img-blog.csdnimg.cn/20201103212248800.png
[20201103212531911.png]: https://img-blog.csdnimg.cn/20201103212531911.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk_size_16_color_FFFFFF_t_70 9]: https://img-blog.csdnimg.cn/20201103212558419.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2E3NzIzMDQ0MTk=,size_16,color_FFFFFF,t_70
[20201103213309826.png]: https://img-blog.csdnimg.cn/20201103213309826.png