KDP数据分析实战：从0到1完成数据实时采集处理到可视化

智领云自主研发的开源轻量级Kubernetes数据平台，即Kubernetes Data Platform (简称KDP)，能够为用户提供在Kubernetes上的一站式云原生数据集成与开发平台。在最新的v1.1.0版本中，用户可借助 KDP 平台上开箱即用的 Airflow、AirByte、Flink、Kafka、MySQL、ClickHouse、Superset 等开源组件快速搭建实时、半实时或批量采集、处理、分析的数据流水线以及可视化报表展示，可视化展示效果如下：

以下我们将介绍一个实时订单数据流水线从数据采集到数据处理，最后到可视化展示的详细建设流程。

1.流水线设计

借助 KDP 平台的开源组件 Airflow、MySQL、Flink、Kafka、ClickHouse、Superset 完成数据实时采集处理及可视化分析，架构如下：

1.1 数据流

直接使用Flink构建实时数仓，由Flink进行清洗加工转换和聚合汇总，将各层结果集写入Kafka中；
ClickHouse从Kafka分别订阅各层数据，将各层数据持久化到ClickHouse中，用于之后的查询分析。

1.2 数据表

本次分析数据基于mock数据，包含数据实时采集处理及可视化分析：

消费者表：customers

字段	字段说明
id	用户ID
name	姓名
age	年龄
gender	性别

订单表：orders

字段	字段说明
order_id	订单ID
order_revenue	订单金额
order_region	下单地区
customer_id	用户ID
create_time	下单时间

1.3 环境说明

在 KDP 页面安装如下组件并完成组件的 QuickStart:

MySQL: 实时数据数据源及 Superset/Airflow 元数据库，安装时需要开启binlog
Kafka: 数据采集sink
Flink: 数据采集及数据处理
ClickHouse: 数据存储
Superset: 数据可视化
Airflow: 作业调度

2. 数据集成与处理

文中使用的账号密码信息请根据实际集群配置进行修改。

2.1 创建MySQL表

2.2 创建 Kafka Topic

进入Kafka broker pod，执行命令创建 Topic，也可以通过Kafka manager 页面创建，以下为进入pod并通过命令行创建的示例：

export BOOTSTRAP="kafka-3-cluster-kafka-0.kafka-3-cluster-kafka-brokers.kdp-data.svc.cluster.local:9092" 


bin/kafka-topics.sh --create \
  --topic ods-order \
  --replication-factor 3 \
  --partitions 10 \
  --bootstrap-server $BOOTSTRAP 


bin/kafka-topics.sh --create \
  --topic ods-customers \
  --replication-factor 3 \
  --partitions 10 \
  --bootstrap-server $BOOTSTRAP


bin/kafka-topics.sh --create \
  --topic dwd-order-customer-valid \
  --replication-factor 3 \
  --partitions 10 \
  --bootstrap-server $BOOTSTRAP


bin/kafka-topics.sh --create \
  --topic dws-agg-by-region \
  --replication-factor 3 \
  --partitions 10 \
  --bootstrap-server $BOOTSTRAP

2.3 创建 ClickHouse 表

进入clickhouse pod，使用`clickhouse-client`执行命令创建表，以下为建表语句：

CREATE DATABASE IF NOT EXISTS kdp_demo;
USE kdp_demo;


-- kafka_dwd_order_customer_valid
CREATE TABLE IF NOT EXISTS kdp_demo.dwd_order_customer_valid (
  order_id Int32,
  order_revenue Float32,
  order_region String,
  create_time DateTime,
  customer_id Int32,
  customer_age Float32,
  customer_name String,
  customer_gender String
) ENGINE = MergeTree()
ORDER BY order_id;


CREATE TABLE kdp_demo.kafka_dwd_order_customer_valid (
  order_id Int32,
  order_revenue Float32,
  order_region String,
  create_time DateTime,
  customer_id Int32,
  customer_age Float32,
  customer_name String,
  customer_gender String
) ENGINE = Kafka
SETTINGS
  kafka_broker_list = 'kafka-3-cluster-kafka-0.kafka-3-cluster-kafka-brokers.kdp-data.svc.cluster.local:9092',
  kafka_topic_list = 'dwd-order-customer-valid',
  kafka_group_name = 'clickhouse_group',
  kafka_format = 'JSONEachRow',
  kafka_row_delimiter = '\n';


CREATE MATERIALIZED VIEW kdp_demo.mv_dwd_order_customer_valid TO kdp_demo.dwd_order_customer_valid AS
SELECT
  order_id,
  order_revenue,
  order_region,
  create_time,
  customer_id,
  customer_age,
  customer_name,
  customer_gender
FROM kdp_demo.kafka_dwd_order_customer_valid;


-- kafka_dws_agg_by_region
CREATE TABLE IF NOT EXISTS kdp_demo.dws_agg_by_region (
  order_region String,
  order_cnt Int64,
  order_total_revenue Float32
) ENGINE = ReplacingMergeTree()
ORDER BY order_region;


CREATE TABLE kdp_demo.kafka_dws_agg_by_region (
  order_region String,
  order_cnt Int64,
  order_total_revenue Float32
) ENGINE = Kafka
SETTINGS
  kafka_broker_list = 'kafka-3-cluster-kafka-0.kafka-3-cluster-kafka-brokers.kdp-data.svc.cluster.local:9092',
  kafka_topic_list = 'dws-agg-by-region',
  kafka_group_name = 'clickhouse_group',
  kafka_format = 'JSONEachRow',
  kafka_row_delimiter = '\n';


CREATE MATERIALIZED VIEW kdp_demo.mv_dws_agg_by_region TO kdp_demo.dws_agg_by_region AS
SELECT
  order_region,
  order_cnt,
  order_total_revenue
FROM kdp_demo.kafka_dws_agg_by_region;

2.4 创建 Flink SQL 作业

2.4.1 SQL部分

CREATE DATABASE IF NOT EXISTS `default_catalog`.`kdp_demo`;


-- create source tables
CREATE TABLE IF NOT EXISTS `default_catalog`.`kdp_demo`.`orders_src`(
    `order_id` INT NOT NULL,
    `order_revenue` FLOAT NOT NULL,
    `order_region` STRING NOT NULL,
    `customer_id` INT NOT NULL,
    `create_time` TIMESTAMP,
    PRIMARY KEY(`order_id`) NOT ENFORCED
) with (
    'connector' = 'mysql-cdc',
    'hostname' = 'kdp-data-mysql',
    'port' = '3306',
    'username' = 'bdos_dba',
    'password' = 'KdpDba!mysql123',
    'database-name' = 'kdp_demo',
    'table-name' = 'orders'
);


CREATE TABLE IF NOT EXISTS `default_catalog`.`kdp_demo`.`customers_src` (
    `id` INT NOT NULL,
    `age` FLOAT NOT NULL,
    `name` STRING NOT NULL,
    `gender` STRING NOT NULL,
    PRIMARY KEY(`id`) NOT ENFORCED
) with (
    'connector' = 'mysql-cdc',
    'hostname' = 'kdp-data-mysql',
    'port' = '3306',
    'username' = 'bdos_dba',
    'password' = 'KdpDba!mysql123',
    'database-name' = 'kdp_demo',
    'table-name' = 'customers'
);


-- create ods dwd and dws tables
CREATE TABLE IF NOT EXISTS `default_catalog`.`kdp_demo`.`ods_order_table` (
    `order_id` INT,
    `order_revenue` FLOAT,
    `order_region` VARCHAR(40),
    `customer_id` INT,
    `create_time` TIMESTAMP,
    PRIMARY KEY (order_id) NOT ENFORCED
) WITH (
    'connector' = 'upsert-kafka',
    'topic' = 'ods-order',
    'properties.bootstrap.servers' = 'kafka-3-cluster-kafka-0.kafka-3-cluster-kafka-brokers.kdp-data.svc.cluster.local:9092',
    'key.format' = 'json',
    'value.format' = 'json'
);


CREATE TABLE IF NOT EXISTS `default_catalog`.`kdp_demo`.`ods_customers_table` (
    `customer_id` INT,
    `customer_age` FLOAT,
    `customer_name` STRING,
    `gender` STRING,
    PRIMARY KEY (customer_id) NOT ENFORCED
) WITH (
    'connector' = 'upsert-kafka',
    'topic' = 'ods-customers',
    'properties.bootstrap.servers' = 'kafka-3-cluster-kafka-0.kafka-3-cluster-kafka-brokers.kdp-data.svc.cluster.local:9092',
    'key.format' = 'json',
    'value.format' = 'json'
);


CREATE TABLE IF NOT EXISTS `default_catalog`.`kdp_demo`.`dwd_order_customer_valid` (
    `order_id` INT,
    `order_revenue` FLOAT,
    `order_region` STRING,
    `create_time` TIMESTAMP,
    `customer_id` INT,
    `customer_age` FLOAT,
    `customer_name` STRING,
    `customer_gender` STRING,
    PRIMARY KEY (order_id) NOT ENFORCED
) WITH (
    'connector' = 'upsert-kafka',
    'topic' = 'dwd-order-customer-valid',
    'properties.bootstrap.servers' = 'kafka-3-cluster-kafka-0.kafka-3-cluster-kafka-brokers.kdp-data.svc.cluster.local:9092',
    'key.format' = 'json',
    'value.format' = 'json'
);


CREATE TABLE IF NOT EXISTS `default_catalog`.`kdp_demo`.`dws_agg_by_region` (
    `order_region` VARCHAR(40),
    `order_cnt` BIGINT,
    `order_total_revenue` FLOAT,
    PRIMARY KEY (order_region) NOT ENFORCED
) WITH (
    'connector' = 'upsert-kafka',
    'topic' = 'dws-agg-by-region',
    'properties.bootstrap.servers' = 'kafka-3-cluster-kafka-0.kafka-3-cluster-kafka-brokers.kdp-data.svc.cluster.local:9092',
    'key.format' = 'json',
    'value.format' = 'json'
);


USE kdp_demo;
-- EXECUTE STATEMENT SET
-- BEGIN
INSERT INTO ods_order_table SELECT * FROM orders_src;
INSERT INTO ods_customers_table SELECT * FROM customers_src;
INSERT INTO
    dwd_order_customer_valid
SELECT
    o.order_id,
    o.order_revenue,
    o.order_region,
    o.create_time,
    c.id as customer_id,
    c.age as customer_age,
    c.name as customer_name,
    c.gender as customer_gender
FROM
    customers_src c
        JOIN orders_src o ON c.id = o.customer_id
WHERE
    c.id <> -1;
INSERT INTO
    dws_agg_by_region
SELECT
    order_region,
    count(*) as order_cnt,
    sum(order_revenue) as order_total_revenue
FROM
    dwd_order_customer_valid
GROUP BY
    order_region;
-- END;

2.4.2 使用 StreamPark 创建 Flink SQL 作业

具体使用参考 StreamPark 文档。

maven 依赖：

<dependency>
    <groupId>com.ververica</groupId>
    <artifactId>flink-sql-connector-mysql-cdc</artifactId>
    <version>3.0.1</version>
</dependency>

2.5 创建 Airflow DAG

2.5.1 DAG 文件部分

import random
from datetime import timedelta
from airflow import DAG
from airflow.operators.bash import BashOperator
from airflow.utils.dates import days_ago




default_args = {
    'owner': 'admin',
    'depends_on_past': False,
    'email_on_failure': False,
    'email_on_retry': False,
    'retries': 1,
}


dag = DAG(
    'kdp_demo_order_data_insert',
    description='Insert into orders by using random data',
    schedule_interval=timedelta(minutes=1),
    start_date=days_ago(1),
    catchup=False,
    tags=['kdp-example'],
)


# MySQL connection info
mysql_host = 'kdp-data-mysql'
mysql_db = 'kdp_demo'
mysql_user = 'bdos_dba'
mysql_password = 'KdpDba!mysql123'
mysql_port = '3306'
cities = ["北京", "上海", "广州", "深圳", "成都", "杭州", "重庆", "武汉", "西安", "苏州", "天津", "南京", "郑州",
          "长沙", "东莞", "青岛", "宁波", "沈阳", "昆明", "合肥", "大连", "厦门", "哈尔滨", "福州", "济南", "温州",
          "佛山", "南昌", "长春", "贵阳", "南宁", "金华", "石家庄", "常州", "泉州", "南通", "太原", "徐州", "嘉兴",
          "乌鲁木齐", "惠州", "珠海", "扬州", "兰州", "烟台", "汕头", "潍坊", "保定", "海口"]
city = random.choice(cities)
consumer_id = random.randint(1, 100)
order_revenue = random.randint(1, 100)
# 插入数据的 BashOperator
insert_data_orders = BashOperator(
    task_id='insert_data_orders',
    bash_command=f'''
    mysql -h {mysql_host} -P {mysql_port} -u {mysql_user} -p{mysql_password} {mysql_db} -e "
    INSERT INTO orders(order_revenue,order_region,customer_id) VALUES({order_revenue},'{city}',{consumer_id});"
    ''',
    dag=dag,
)
insert_data_orders

2.5.2 DAG 说明及执行

当前Airflow安装时，需要指定可访问的git 仓库地址，因此需要将 Airflow DAG 提交到 Git 仓库中。每分钟向orders表插入一条数据。

2.6 数据验证

使用ClickHouse验证数据：

（1）进入ClickHouse客户端

clickhouse-client 
# default pass: ckdba.123

（2）执行查询

SELECT * FROM kdp_demo.dwd_order_customer_valid;
SELECT count(*) FROM kdp_demo.dwd_order_customer_valid;

（3）对比验证MySQL中数据是否一致

select count(*) from kdp_demo.orders;

3. 数据可视化

在2.6中数据验证通过后，可以通过Superset进行数据可视化展示。使用账号`admin/admin`登录Superset页面（注意添加本地 Host 解析）：http://superset-kdp-data.kdp-e2e.io

3.1 创建图表

导入我们制作好的图表：

下载面板：https://gitee.com/linktime-cloud/example-datasets/raw/main/superset/dashboard_export_20240607T100739.zip
导入面板

（1）选择下载的文件导入

（2）输入 ClickHouse 的用户`default`的默认密码`ckdba.123`：

3.2 效果展示

最终的实时订单数据图表展示如下，随着订单数据的更新，图表中的数据也会实时更新：

快速体验

🚀GitHub项目：

https://github.com/linktimecloud/kubernetes-data-platform

欢迎您参与开源社区的建设🤝

- FIN -

更多精彩推荐

我们开源啦！一键部署免费使用！Kubernetes上直接运行大数据平台！
开源 KDP v1.1.0 版本正式发布，新增数据集成开发应用场景
在 KubeSphere 上快速安装和使用 KDP 云原生数据平台
在 Rancher 上快速安装和使用 KDP 云原生数据平台