最近调研 Doris,设想了针对行为分析、用户分析的应用场景,看看 Doris 匹配程度。
自动更新保持保持数据的一致性,减少维护成本。
# 约束
Centos >= 7.1
Java >= 1.8
GCC >= 4.8.2
# 文件句柄
vi /etc/security/limits.conf
* soft nofile 65536
* hard nofile 65536
# 其他
时钟同步
关闭交换分区(swap)
Linux(ext4)
复制代码
# 1. 下载FE
wget --no-check-certificate https://mirrors.tuna.tsinghua.edu.cn/apache/doris/1.1/1.1.4-rc01/apache-doris-fe-1.1.4-bin.tar.gz
# 2. 解压
tar -xvf apache-doris-fe-1.1.4-bin.tar.gz
# 3. 启动
cd apache-doris-fe-1.1.4-bin && bin/start_fe.sh --daemon
# 4. 安装MySql Client
wget http://repo.mysql.com/mysql57-community-release-el7-8.noarch.rpm
rpm -ivh mysql57-community-release-el7-8.noarch.rpm
rpm --import https://repo.mysql.com/RPM-GPG-KEY-mysql-2022
yum install mysql-community-client
# 5. 连接FE
mysql -h 172.30.144.1 -P 9030 -uroot
# 6. 查看FE节点状态
SHOW PROC '/frontends'G;
复制代码
# 1. 下载
wget --no-check-certificate https://mirrors.tuna.tsinghua.edu.cn/apache/doris/1.1/1.1.4-rc01/apache-doris-be-1.1.4-bin-x86_64.tar.gz
# 2. 解压
tar -xvf apache-doris-be-1.1.4-bin-x86_64.tar.gz
# 3. 启动:同样启动其他节点
cd apache-doris-be-1.1.4-bin-x86_64 && bin/start_be.sh --daemon
# 4. 连接FE
mysql -h 172.30.144.1 -P 9030 -uroot
# 5. 添加BE
ALTER SYSTEM ADD BACKEND "172.30.144.2:9050";
ALTER SYSTEM ADD BACKEND "172.30.144.3:9050";
ALTER SYSTEM ADD BACKEND "172.30.144.4:9050";
# 6. 查看BE状态
SHOW PROC '/backends'G
复制代码
# 1. 启动
cd /root/apache-doris-fe-1.1.4-bin/apache_hdfs_broker
bin/start_broker.sh --daemon
# 2. 连接FE
mysql -h 172.30.144.1 -P 9030 -uroot
# 3. 添加Broker
ALTER SYSTEM ADD BROKER broker_name "172.30.144.5:8000";
# 4. 查看Broker节点
SHOW PROC "/brokers"G
复制代码
数据写入时,按照维度列对其他指标列进行聚合操作,操作结果:多条数据如果所有维度列相同,那么会对所有指标列进行聚合。
-- 1. 创建表
CREATE TABLE IF NOT EXISTS zgg.user
(
`city` VARCHAR(20) COMMENT "城市",
`age` SMALLINT COMMENT "年龄",
`sex` TINYINT COMMENT "性别",
`pv` BIGINT SUM DEFAULT "0" COMMENT "pv",
`min_time` INT MIN DEFAULT "0" COMMENT "最小停留时间",
`max_time` INT MAX DEFAULT "0" COMMENT "最大停留时间"
)
AGGREGATE KEY(`city`, `age`, `sex`)
DISTRIBUTED BY HASH(`city`) BUCKETS 3;
-- 2. 写入数据
INSERT INTO zgg.user VALUES('北京', 25, 1, 1, 50, 100);
INSERT INTO zgg.user VALUES('北京', 20, 0, 1, 150, 300);
复制代码
-- 3. 写入相同数据
INSERT INTO zgg.user VALUES('北京', 25, 1, 1, 15, 3100);
复制代码
通过定义主键保证数据的唯一性,Unique 模型简化了数据导入流程,能够更好地支撑实时和频繁更新的场景。
-- 1. 创建表
CREATE TABLE IF NOT EXISTS zgg.user_unique
(
`user_id` LARGEINT NOT NULL COMMENT "用户id",
`city` VARCHAR(20) COMMENT "城市",
`age` SMALLINT COMMENT "年龄",
`sex` TINYINT COMMENT "性别"
)
UNIQUE KEY(`user_id`)
DISTRIBUTED BY HASH(`user_id`) BUCKETS 3;
-- 2. 写入数据
INSERT INTO zgg.user_unique VALUES(1, '北京', 25, 1);
INSERT INTO zgg.user_unique VALUES(2, '上海', 35, 0);
复制代码
-- 3. 写入主键相同数据
INSERT INTO zgg.user_unique VALUES(1, '沈阳', 15, 1);
复制代码
用于存储原始数据,允许重复数据
-- 1. 创建表
CREATE TABLE IF NOT EXISTS zgg.user_duplicate
(
`user_id` LARGEINT NOT NULL COMMENT "用户id",
`city` VARCHAR(20) COMMENT "城市",
`age` SMALLINT COMMENT "年龄",
`sex` TINYINT COMMENT "性别"
)
DUPLICATE KEY(`user_id`)
DISTRIBUTED BY HASH(`user_id`) BUCKETS 3;
-- 2. 写入数据
INSERT INTO zgg.user_duplicate VALUES(1, '北京', 25, 1);
INSERT INTO zgg.user_duplicate VALUES(2, '上海', 35, 0);
复制代码
-- 3. 写入主键相同数据
INSERT INTO zgg.user_duplicate VALUES(1, '沈阳', 15, 1);
复制代码
步骤 1:使用 Duplicate 模型存储原始数据
CREATE TABLE IF NOT EXISTS zgg.user_all
(
`date` DATE NOT NULL COMMENT "时间",
`page` VARCHAR(20) NOT NULL COMMENT '页面',
`user_id` LARGEINT NOT NULL COMMENT "用户id"
)
DUPLICATE KEY(`date`, `page`, `user_id`)
DISTRIBUTED BY HASH(`date`) BUCKETS 3;
-- 插入测试数据
INSERT INTO zgg.user_all VALUES('2022-11-18', 'login', 1);
INSERT INTO zgg.user_all VALUES('2022-11-18','login', 2);
INSERT INTO zgg.user_all VALUES('2022-11-18','order', 1);
INSERT INTO zgg.user_all VALUES('2022-11-18','order', 1);
INSERT INTO zgg.user_all VALUES('2022-11-18','order', 2);
INSERT INTO zgg.user_all VALUES('2022-11-18','pay', 1);
INSERT INTO zgg.user_all VALUES('2022-11-18','pay', 1);
复制代码
步骤 2:使用聚合模型计算分页面 PV
-- 1. 创建表
CREATE TABLE IF NOT EXISTS zgg.user_pv
(
`date` DATE NOT NULL COMMENT "时间",
`page` VARCHAR(20) NOT NULL COMMENT '页面',
`pv` BIGINT SUM DEFAULT "0" COMMENT "pv"
)
AGGREGATE KEY(`date`, `page`)
DISTRIBUTED BY HASH(`date`) BUCKETS 3;
-- 2. 使用INSERT INTO table SELECT ...
INSERT INTO zgg.user_pv SELECT `date`, `page`, 1 FROM zgg.user_all;
复制代码
步骤 3:使用 Unique 模型计算分页面 UV
-- 1. 创建表
CREATE TABLE IF NOT EXISTS zgg.user_uv
(
`date` DATE NOT NULL COMMENT "时间",
`page` VARCHAR(20) NOT NULL COMMENT '页面',
`user_id` LARGEINT NOT NULL COMMENT "用户id"
)
UNIQUE KEY(`date`, `page`, `user_id`)
DISTRIBUTED BY HASH(`date`) BUCKETS 3;
-- 2. 使用INSERT INTO table SELECT ...
INSERT INTO zgg.user_uv SELECT `date`, `page`, `user_id` FROM zgg.user_all;
-- 3. 使用COUNT查询UV
SELECT date, page, count(user_id) uv
FROM zgg.user_uv
GROUP BY date, page;
复制代码
使用 Unique 模型
-- 1. 注册表
CREATE TABLE IF NOT EXISTS zgg.user_register
(
`date` DATE NOT NULL COMMENT "时间",
`user_id` LARGEINT NOT NULL COMMENT "用户id"
)
UNIQUE KEY(`date`, `user_id`)
DISTRIBUTED BY HASH(`date`) BUCKETS 3;
-- 2. 订单表
CREATE TABLE IF NOT EXISTS zgg.user_order
(
`date` DATE NOT NULL COMMENT "时间",
`user_id` LARGEINT NOT NULL COMMENT "用户id"
)
UNIQUE KEY(`date`, `user_id`)
DISTRIBUTED BY HASH(`date`) BUCKETS 3;
-- 3. 获取近7天注册总用户
SELECT COUNT(t1.user_id) cnt
FROM zgg.user_register t1
AND t1.`date` BETWEEN '2022-11-12' AND '2022-11-18';
-- 4. 获取近7天注册用户中下订单的用户
SELECT COUNT(t2.user_id)
FROM zgg.user_register t1, zgg.user_order t2
WHERE t1.user_id = t2.user_id
AND t1.`date` BETWEEN '2022-11-12' AND '2022-11-18';
复制代码
使用聚合模型 + Bitmap
-- 1. 创建表
CREATE TABLE IF NOT EXISTS zgg.user_funnel
(
`date` DATE NOT NULL COMMENT "时间",
`page` VARCHAR(20) NOT NULL COMMENT '页面',
`user_id` BITMAP BITMAP_UNION NULL COMMENT "用户id"
)
AGGREGATE KEY(`date`, `page`)
DISTRIBUTED BY HASH(`date`) BUCKETS 3;
-- 2. 写入模拟数据
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'register', to_bitmap(1));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'register', to_bitmap(2));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'register', to_bitmap(3));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'register', to_bitmap(4));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'shopping', to_bitmap(1));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'shopping', to_bitmap(2));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'shopping', to_bitmap(3));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'order', to_bitmap(1));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'order', to_bitmap(2));
INSERT INTO zgg.user_funnel VALUES('2022-11-18', 'pay', to_bitmap(1));
-- 3. 通过bitmap函数:bitmap_count和bitmap_union组合以及intersect_count
SELECT 'register' page, bitmap_count(bitmap_union(user_id)) cnt
FROM zgg.user_funnel
UNION ALL
-- register -> shopping
SELECT 'register -> shopping' page, intersect_count(user_id, page, 'register', 'shopping') cnt
FROM zgg.user_funnel
UNION ALL
-- register -> shopping -> order
SELECT 'register -> shopping -> order' page, intersect_count(user_id, page, 'register', 'shopping', 'order') cnt
FROM zgg.user_funnel
UNION ALL
-- register -> shopping -> order -> 'pay'
SELECT 'register -> shopping -> order-> pay' page, intersect_count(user_id, page, 'register', 'shopping', 'order', 'pay') cnt
FROM zgg.user_funnel;
复制代码
其余场景使用 Doris 基本都可以适配,满足功能需求后,接下来会着重进行故障测试和性能测试。
页面更新:2024-04-20
本站资料均由网友自行发布提供,仅用于学习交流。如有版权问题,请与我联系,QQ:4156828
© CopyRight 2008-2024 All Rights Reserved. Powered By bs178.com 闽ICP备11008920号-3
闽公网安备35020302034844号