服务器之家:专注于VPS、云服务器配置技术及软件下载分享
分类导航

Mysql|Sql Server|Oracle|Redis|MongoDB|PostgreSQL|Sqlite|DB2|mariadb|Access|数据库技术|

服务器之家 - 数据库 - 数据库技术 - Clickhouse系列之整合Hive数据仓库示例详解

Clickhouse系列之整合Hive数据仓库示例详解

2022-12-05 16:22阿杰笔记 数据库技术

这篇文章主要为大家介绍了Clickhouse系列之整合Hive数据仓库示例详解,有需要的朋友可以借鉴参考下,希望能够有所帮助,祝大家多多进步,早日升职加薪

前言

什么是Hive? Apache Hive 数据仓库软件便于使用SQL读取、写入和管理驻留在分布式存储中的大型数据集。结构可以投射到已存储的数据上。提供了一个命令行工具和JDBC驱动程序,用于将用户连接到Hive。

Hive引擎允许您对HDFS配置单元表执行SELECT查询。目前支持如下输入格式:

  • 文本:仅支持简单标量列类型,二进制除外;
  • ORC:支持除char以外的简单标量列类型;仅支持数组等复杂类型;
  • parquet:支持所有简单的标量列类型;仅支持数组等复杂类型。

正文

创建Hive引擎表详细信息以及参数详解

?
1
2
3
4
5
6
7
CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] 
name1 [type1] [ALIAS expr1], 
name2 [type2] [ALIAS expr2], 
... 
) ENGINE = Hive('thrift://host:port', 'database', 'table'); 
PARTITION BY expr

表结构可以与原始配置单元表结构不同:

  • 列名应该与原始配置单元表中的列名相同(推荐列名相同处理),但您可以只使用其中的一些列,并且可以按任何顺序使用,也可以使用从其他列计算的一些别名列。
  • 列类型应与原始配置单元表中的列类型相同
  • 按表达式划分应该与原始Hive表一致,按表达式划分中的列应该在表结构中。

引擎参数:

  • thrift://host:port-配置单元元存储地址
  • database—远程数据库名称。
  • table—远程表名称。

实战案例

为远程文件系统启用本地缓存。通过官方的基准测试表明,使用缓存的速度快了近两倍。在使用缓存之前,将其添加到config.xml

?
1
2
3
4
5
6
<local_cache_for_remote_fs>
    <enable>true</enable>
    <root_dir>local_cache</root_dir>
    <limit_size>559096952</limit_size>
    <bytes_read_before_flush>1048576</bytes_read_before_flush>
</local_cache_for_remote_fs>

参数详解:

  • enable:ClickHouse将在启动后维护远程文件系统(HDFS)的本地缓存(如果为true)。
  • root_dir:必需。用于存储远程文件系统的本地缓存文件的根目录。
  • limit_size:必填。本地缓存文件的最大大小(字节)。
  • bytes_read_before_flush:从远程文件系统下载文件时,在刷新到本地文件系统之前控制字节数。默认值为1MB。

尽管ClickHouse在启用远程文件系统本地缓存的情况下启动时,我们仍然可以选择不使用其查询中设置为use_local_cache_for_remote_fs=0的缓存。use_local_cache_for_remote_fs默认为false

ORC数据格式

  • Hive创建ORC数据格式表
?
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
CREATE TABLE `test`.`test_orc`( 
`f_tinyint` tinyint, 
`f_smallint` smallint
`f_int` int
`f_integer` int
`f_bigint` bigint
`f_float` float
`f_double` double
`f_decimal` decimal(10,0), 
`f_timestamp` timestamp
`f_date` date
`f_string` string, 
`f_varchar` varchar(100), 
`f_bool` boolean, 
`f_binary` binary
`f_array_int` array<int>, 
`f_array_string` array<string>, 
`f_array_float` array<float>, 
`f_array_array_int` array<array<int>>, 
`f_array_array_string` array<array<string>>, 
`f_array_array_float` array<array<float>>) 
PARTITIONED BY
`day` string) 
ROW FORMAT SERDE 
'org.apache.hadoop.hive.ql.io.orc.OrcSerde'
STORED AS INPUTFORMAT 
'org.apache.hadoop.hive.ql.io.orc.OrcInputFormat'
OUTPUTFORMAT 
'org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat'
LOCATION 
'hdfs://testcluster/data/hive/test.db/test_orc'
?
1
insert into test.test_orc partition(day='2021-09-18') select 1, 2, 3, 4, 5, 6.11, 7.22, 8.333, current_timestamp(), current_date(), 'hello world', 'hello world', 'hello world', true, 'hello world', array(1, 2, 3), array('hello world', 'hello world'), array(float(1.1), float(1.2)), array(array(1, 2), array(3, 4)), array(array('a', 'b'), array('c', 'd')), array(array(float(1.11), float(2.22)), array(float(3.33), float(4.44)));
  • Clickhouse创建Hive表引擎
?
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
CREATE TABLE test.test_orc
(
    `f_tinyint` Int8,
    `f_smallint` Int16,
    `f_int` Int32,
    `f_integer` Int32,
    `f_bigint` Int64,
    `f_float` Float32,
    `f_double` Float64,
    `f_decimal` Float64,
    `f_timestamp` DateTime,
    `f_date` Date,
    `f_string` String,
    `f_varchar` String,
    `f_bool` Bool,
    `f_binary` String,
    `f_array_int` Array(Int32),
    `f_array_string` Array(String),
    `f_array_float` Array(Float32),
    `f_array_array_int` Array(Array(Int32)),
    `f_array_array_string` Array(Array(String)),
    `f_array_array_float` Array(Array(Float32)),
    `day` String
)
ENGINE = Hive('thrift://202.168.117.26:9083', 'test', 'test_orc')
PARTITION BY day
  • 通过Clickhouse查询Hive数据
?
1
SELECT * FROM test.test_orc settings input_format_orc_allow_missing_columns = 1\G

Parquet数据格式

  • Hive创建Parquet数据格式表
?
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
CREATE TABLE `test`.`test_parquet`( 
`f_tinyint` tinyint, 
`f_smallint` smallint
`f_int` int
`f_integer` int
`f_bigint` bigint
`f_float` float
`f_double` double
`f_decimal` decimal(10,0), 
`f_timestamp` timestamp
`f_date` date
`f_string` string, 
`f_varchar` varchar(100), 
`f_char` char(100), 
`f_bool` boolean, 
`f_binary` binary
`f_array_int` array<int>, 
`f_array_string` array<string>, 
`f_array_float` array<float>, 
`f_array_array_int` array<array<int>>, 
`f_array_array_string` array<array<string>>, 
`f_array_array_float` array<array<float>>) 
PARTITIONED BY
`day` string) 
ROW FORMAT SERDE 
'org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe'
STORED AS INPUTFORMAT 
'org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat'
OUTPUTFORMAT 
'org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat'
LOCATION 
'hdfs://testcluster/data/hive/test.db/test_parquet'
?
1
insert into test.test_parquet partition(day='2021-09-18') select 1, 2, 3, 4, 5, 6.11, 7.22, 8.333, current_timestamp(), current_date(), 'hello world', 'hello world', 'hello world', true, 'hello world', array(1, 2, 3), array('hello world', 'hello world'), array(float(1.1), float(1.2)), array(array(1, 2), array(3, 4)), array(array('a', 'b'), array('c', 'd')), array(array(float(1.11), float(2.22)), array(float(3.33), float(4.44)));
  • Clickhouse创建Hive表引擎
?
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
CREATE TABLE test.test_parquet 
`f_tinyint` Int8, 
`f_smallint` Int16, 
`f_int` Int32, 
`f_integer` Int32, 
`f_bigint` Int64, 
`f_float` Float32, 
`f_double` Float64, 
`f_decimal` Float64, 
`f_timestamp` DateTime, 
`f_date` Date
`f_string` String, 
`f_varchar` String, 
`f_char` String, 
`f_bool` Bool, 
`f_binary` String, 
`f_array_int` Array(Int32), 
`f_array_string` Array(String), 
`f_array_float` Array(Float32), 
`f_array_array_int` Array(Array(Int32)), 
`f_array_array_string` Array(Array(String)), 
`f_array_array_float` Array(Array(Float32)), 
`day` String 
ENGINE = Hive('thrift://localhost:9083', 'test', 'test_parquet'
PARTITION BY day
  • 通过Clickhouse查询Hive数据
?
1
SELECT * FROM test.test_parquet settings input_format_parquet_allow_missing_columns = 1\G

TextFile数据格式

  • Hive创建TextFile数据格式表
?
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
CREATE TABLE `test`.`test_text`( 
`f_tinyint` tinyint, 
`f_smallint` smallint
`f_int` int
`f_integer` int
`f_bigint` bigint
`f_float` float
`f_double` double
`f_decimal` decimal(10,0), 
`f_timestamp` timestamp
`f_date` date
`f_string` string, 
`f_varchar` varchar(100), 
`f_char` char(100), 
`f_bool` boolean, 
`f_binary` binary
`f_array_int` array<int>, 
`f_array_string` array<string>, 
`f_array_float` array<float>, 
`f_array_array_int` array<array<int>>, 
`f_array_array_string` array<array<string>>, 
`f_array_array_float` array<array<float>>) 
PARTITIONED BY
`day` string) 
ROW FORMAT SERDE 
'org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe'
STORED AS INPUTFORMAT 
'org.apache.hadoop.mapred.TextInputFormat'
OUTPUTFORMAT 
'org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat'
LOCATION 
'hdfs://testcluster/data/hive/test.db/test_text'
?
1
insert into test.test_text partition(day='2021-09-18') select 1, 2, 3, 4, 5, 6.11, 7.22, 8.333, current_timestamp(), current_date(), 'hello world', 'hello world', 'hello world', true, 'hello world', array(1, 2, 3), array('hello world', 'hello world'), array(float(1.1), float(1.2)), array(array(1, 2), array(3, 4)), array(array('a', 'b'), array('c', 'd')), array(array(float(1.11), float(2.22)), array(float(3.33), float(4.44)));
  • Clickhouse创建Hive表引擎
?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
CREATE TABLE test.test_text 
`f_tinyint` Int8, 
`f_smallint` Int16, 
`f_int` Int32, 
`f_integer` Int32, 
`f_bigint` Int64, 
`f_float` Float32, 
`f_double` Float64, 
`f_decimal` Float64, 
`f_timestamp` DateTime, 
`f_date` Date
`f_string` String, 
`f_varchar` String, 
`f_char` String, 
`f_bool` Bool, 
`day` String 
ENGINE = Hive('thrift://localhost:9083', 'test', 'test_text'
PARTITION BY day
  • 通过Clickhouse查询Hive数据
?
1
SELECT * FROM test.test_text settings input_format_skip_unknown_fields = 1, input_format_with_names_use_header = 1, date_time_input_format = 'best_effort'\G

总结

本节主要讲解了Clickhouse整合Hive数仓,利用了Hive引擎并通过thrift方式去连接,需要注意这种连接参数的设置以及代表意义。另外,这个过程我们需要注意的是,推荐开启缓存,这样查询速度会快很多。与此同时,也对Hive常用的三种数据类型ORC,Parquet,TextFile进行了一个实战案例操作,更多关于Clickhouse整合Hive数据仓库的资料请关注服务器之家其它相关文章!

原文链接:https://juejin.cn/post/7153507694140194823

延伸 · 阅读

精彩推荐