背景

ByteHouse 是火山引擎旗下基于开源 ClickHouse 的企业级分析型数据库，是一个同时支持实时和离线导入的自助数据分析平台，能够对 PB 级海量数据进行高效分析。
本文将介绍如何在 E-MapReduce（EMR） 集群提交 Flink SQL 和 Flink jar 任务，将数据写入到 ByteHouse 集群的方法。

EMR Flink 数据写入ByteHouse（云数仓版）

前提条件

• 已创建火山引擎 EMR 集群。具体操作，请参见 创建集群。
• 已创建火山引擎 ByteHouse 集群。具体操作，请参见 ByteHouse 快速入门-火山引擎。

准备工作

• 生成访问密钥，在火山引擎的 密钥管理 页面，查找对应用户的访问秘钥（Access Key ID 和 Secret Access Key）
  
• 向 ByteHouse 写数据，是通过 ByteHouse Gateway 实现的。具体方式为在使用过程中将参数 Region ，根据使用场景设置为不同的值 。同时需要 EMR 集群的各个节点能够与之进行通信，当前有以下两种方式：
  • 设置 Region 为 VOLCANO，给 EMR 集群的每个节点绑定一个公网 IP；
  • ByteHouse Gateway 也支持火山引擎内网访问方式，需要 ByteHouse 侧给 EMR 集群加白，可 联系客服 进行操作。

Flink SQL 集群执行

运行 Flink SQL client 时根据如下路径指定 jar。

cd /usr/lib/emr/current
./bin/sql-client.sh --jar connectors/flink-connector-bytehouse-cdw-assembly-x.x.x-x.x.jar

可运行如下 SQL，进行测试运行。

CREATE  TABLE random_source (
            test_key   STRING,
            test_value BIGINT,
            ts         BIGINT
        )
        WITH (
            'connector' = 'datagen',
            'rows-per-second' = '1'
        );

CREATE  TABLE cnch_table (test_key STRING, test_value BIGINT, ts BIGINT)
        WITH (
            'connector' = 'bytehouse-cdw',
            'database' = '<此处填写 ByteHouse 库名>',
            'table-name' = '<此处填写 ByteHouse 表名>',
            'bytehouse.gateway.region' = 'VOLCANO',
            'bytehouse.gateway.access-key-id' = '<此处填写用户实际的 AK>',
            'bytehouse.gateway.secret-key' = '<此处填写用户实际的 SK>'
        );

INSERT INTO cnch_table
SELECT  *
FROM    random_source;

Flink JAR 包作业集群提交

下载对应版本的 Connector

访问 ByteHouse Connector 下载地址，选择对应版本目录下的文件进行下载。
下载后的文件命名格式为：flink-connector-bytehouse-cdw-assembly-#.#.#-#.#.jar。

导入依赖

在本地 Maven 项目的 pom.xml 文件中添加以下配置以导入对应依赖，其中 flink-cnch-connector 中安装至本地 maven 仓库。

mvn install:install-file   -Dfile=/xxx/xxx/flink-connector-bytehouse-cdw-assembly-1.11.4-1.15.jar  -DgroupId=com.bytedance  -DartifactId=flink-cnch-connector -Dversion=1.0.0  -Dpackaging=jar

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0"
  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
  xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
  <modelVersion>4.0.0</modelVersion>

  <groupId>org.example</groupId>
  <artifactId>flinkTest</artifactId>
  <version>1.0-SNAPSHOT</version>

  <properties>
    <java.version>1.8</java.version>
    <maven.compiler.source>8</maven.compiler.source>
    <maven.compiler.target>8</maven.compiler.target>
    <flink.version>1.11.3</flink.version>
    <scala.version>2.11</scala.version>
  </properties>


  <dependencies>
    <dependency>
      <groupId>org.apache.flink</groupId>
      <artifactId>flink-core</artifactId>
      <version>${flink.version}</version>
    </dependency>

    <dependency>
      <groupId>org.apache.flink</groupId>
      <artifactId>flink-java</artifactId>
      <version>${flink.version}</version>
    </dependency>

    <dependency>
      <groupId>org.apache.flink</groupId>
      <artifactId>flink-clients_${scala.version}</artifactId>
      <version>${flink.version}</version>
    </dependency>

    <dependency>
      <groupId>org.apache.flink</groupId>
      <artifactId>flink-table-api-java-bridge_${scala.version}</artifactId>
      <version>${flink.version}</version>
    </dependency>

    <dependency>
      <groupId>com.bytedance</groupId>
      <artifactId>flink-cnch-connector</artifactId>
      <version>1.0</version>
   </dependencies>

</project>

DummyRowData.java 源数据生成样例

/**
  * Synthetic {@code  RowData} generator mimicking the feed of crime cases reported by Neighbourhood
  * Police Centres (NPCs) in Singapore.
  */
public class DummyRowDataSource extends RichParallelSourceFunction<RowData> {

    private static final AtomicLong pullCount = new AtomicLong();

    static Map<String, Timer> map  = new ConcurrentHashMap<>(2, 0.9f, 1);

    private final List<String> offences = Arrays.asList("Unlicensed Moneylending", "Harassment");

    private final AtomicInteger caseNo = new AtomicInteger();

    private volatile boolean cancelled = false;

    private Random random;

    @Override
    public void open(Configuration parameters) throws Exception {
        super.open(parameters);
        random = new Random();

        map.computeIfAbsent(
                "holder",
                s -> {
                    final Timer timer = new Timer("RandomStringSource ", true);
                    timer.schedule(
                            new TimerTask() {
                                @Override
                                public void run() {
                                    System.out.printf("source is pulled %s times\n", pullCount.get());
                                }
                            },
                            5000,
                            5000);
                    return timer;
                });
    }

    @Override
    public void run(SourceContext<RowData> ctx) throws Exception {
        while (!cancelled) {
            Thread.sleep(random.nextInt(10) + 5);
            synchronized (ctx.getCheckpointLock()) {
                final GenericRowData genericRowData = new GenericRowData(RowKind.INSERT, 4);
                genericRowData.setField(0, RowDataConversion.fieldDataOf(2000 + random.nextInt(20)));
                genericRowData.setField(1, RowDataConversion.fieldDataOf(generateRandomWord(4)));
                genericRowData.setField(2, RowDataConversion.fieldDataOf(randomOffences()));
                genericRowData.setField(3, RowDataConversion.fieldDataOf(caseNo.incrementAndGet()));

                ctx.collect(genericRowData);
                pullCount.incrementAndGet();
            }
        }
    }

    @Override
    public void cancel() {
        cancelled = true;
    }

    private String generateRandomWord(int wordLength) {
        StringBuilder sb = new StringBuilder(wordLength);
        for (int i = 0; i < wordLength; i++) { // For each letter in the word
            char tmp = (char) ('a' + random.nextInt('z' - 'a')); // Generate a letter between a and z
            sb.append(tmp); // Add it to the String
        }
        return sb.toString();
    }

    private String randomOffences() {
        return offences.get(random.nextInt(2));
    }
}

CnchSinkDataStreamExample.java 数据 sink 样例

根据提示，填写用户实际的 AK，SK，ByteHouse 的库名及表名。

public class CnchSinkDataStreamExample {

    public static void main(String[] args) throws Exception {
        final String region = "VOLCANO";
        final String ak = "<此处填写用户实际的 AK>";
        final String sk = "<此处填写用户实际的 SK>";
        final String dbName = "<此处填写 ByteHouse 库名>";
        final String tableName = "<此处填写 ByteHouse 表名>";

        // create env
        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        // add source
        DataStream<RowData> dataStream =
                env.addSource(new DummyRowDataSource()).returns(TypeInformation.of(RowData.class));

        List<TableColumn> columns =
                Arrays.asList(
                        TableColumn.of("year", DataTypes.INT()),
                        TableColumn.of("npc", DataTypes.STRING()),
                        TableColumn.of("offence", DataTypes.STRING()),
                        TableColumn.of("case_no", DataTypes.INT()));

        try (AbstractClickHouseSinkFunction<RowData, InsertBatch<RowData>, ?> cnchSink =
                new CnchSinkFunctionBuilder(dbName, tableName)
                        .withSchema(columns)
                        .withGatewayConnection(region)
                        .withGatewayCredentials(ak, sk)
                        .withFlushInterval(Duration.ofSeconds(1))
                        .build()) {
            // add sink
            dataStream.addSink(cnchSink);

            // trigger pipeline
            env.execute(CnchSinkDataStreamExample.class.getSimpleName());
        }
    }
}

编译项目生成可运行 jar

为了减少潜在的包冲突文件，建议用户打 fat jar，集成相关依赖。
在项目 pom.xml 文件中，添加如下 build 方式，以生成 fat jar。

<build>
  <finalName>fat-jar-example</finalName>
  <plugins>
    <plugin>
      <groupId>org.apache.maven.plugins</groupId>
      <artifactId>maven-assembly-plugin</artifactId>
      <version>3.4.2</version>
      <configuration>
        <descriptorRefs>
          <descriptorRef>jar-with-dependencies</descriptorRef>
        </descriptorRefs>
      </configuration>
      <executions>
        <execution>
          <id>assemble-all</id>
          <phase>package</phase>
          <goals>
            <goal>single</goal>
          </goals>
        </execution>
      </executions>
    </plugin>
  </plugins>
</build>

上传 jar 包到 EMR 集群

可通过 scp 指令。

scp xxxx.jar root@xxx.xxx.xxx.xxx:/{path}/

执行任务

提交任务的命令：

/usr/lib/emr/current/flink/bin/flink run-application -t yarn-application <JAR 包文件名>

Yarn ResourceManager UI 查看运行状态

Yarn ResourceManager UI 访问方式，可参考访问链接 E-MapReduce-火山引擎。

EMR Flink 数据写入ByteHouse（企业版）

前提条件

• 已创建火山引擎 EMR 集群（集群版本需高于3.8.0）。具体操作，请参见创建集群。
• 已创建火山引擎ByteHouse(企业版)，具体操作，请参见ByteHouse企业版快速入门-火山引擎。

准备工作

在创建的ByteHouse 企业版集群建表

获取用户认证信息

Flink SQL 集群执行

运行Flink SQL client时根据如下路径指定jar。

cd /usr/lib/emr/current
./bin/sql-client.sh --jar connectors/[flink-sql-connector-bytehouse-ce](https://maven.byted.org/repository/releases/com/bytedance/bytehouse/flink-sql-connector-bytehouse-ce_2.12/1.25.4-1.16/flink-sql-connector-bytehouse-ce_2.12-1.25.4-1.16.jar)-x.x.x-x.x.jar

可运行如下SQL，进行测试运行。

CREATE  TABLE random_source (
            test_key   STRING,
            test_value BIGINT,
            ts         BIGINT
        )
        WITH (
            'connector' = 'datagen',
            'rows-per-second' = '1'
        );



CREATE  TABLE cnch_table (test_key STRING, test_value BIGINT, ts BIGINT)
        WITH (
            'connector' = 'bytehouse-ce',
            'clickhouse.shard-discovery.kind' = 'CE_GATEWAY',
            'bytehouse.ce.gateway.host' = '7301548592783020297.bytehouse-ce.ivolces.com',
            'bytehouse.ce.gateway.port' = '8123',            
            'sink.enable-upsert' = 'false',
            'clickhouse.cluster' = 'bytehouse_cluster_enct', -- bytehouse集群名称
            'database' = 'default', -- 目标数据库
            'table-name' = 'zp_test', -- 目标表 注意是local表：{table_name}_local
            'username' = 'xxxx@bytedance.com', -- 2.3步获取的用户名
            'password' = 'xxx' -- 2.3步获取的bytehouse密码
        );

INSERT INTO cnch_table SELECT  * FROM    random_source;

Flink JAR 包作业集群提交

下载 Flink ByteHouse CE 的connector

flink-sql-connector-bytehouse-ce_2.12-1.25.4-1.16.jar

36.17MB

导入依赖

在本地 Maven 项目的 pom.xml 文件中添加以下配置以导入对应依赖，其中 flink-sql-connector-bytehouse-ce 中安装至本地maven仓库。

在本地 Maven 项目的 pom.xml 文件中添加以下配置以导入对应依赖，其中 flink-sql-connector-bytehouse-ce 中安装至本地maven仓库
mvn install:install-file   -Dfile=/xxx/xxx/flink-sql-connector-bytehouse-ce_2.12-1.25.4-1.16.jar -DgroupId=com.bytedance.bytehouse -DartifactId=flink-sql-connector-bytehouse-ce_2.12 -Dversion=1.25.4-1.16 -Dpackaging=jar

<project xmlns="http://maven.apache.org/POM/4.0.0"
         xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
         xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>

    <groupId>example.flink</groupId>
    <artifactId>bytehouse-ce</artifactId>
    <version>0.0.1-SNAPSHOT</version>

    <dependencies>
        <dependency>
            <groupId>com.bytedance.bytehouse</groupId>
            <artifactId>flink-sql-connector-bytehouse-ce_2.12</artifactId>
            <version>1.25.4-1.16</version>
        </dependency>
        <dependency>
            <groupId>org.apache.flink</groupId>
            <artifactId>flink-table-planner_2.12</artifactId>
            <version>1.16.2</version>
            <scope>provided</scope>
        </dependency>
    </dependencies>
    <build>
        <plugins>
            <plugin>
                <groupId>org.apache.maven.plugins</groupId>
                <artifactId>maven-compiler-plugin</artifactId>
                <configuration>
                    <source>8</source>
                    <target>8</target>
                </configuration>
            </plugin>
        </plugins>
    </build>

</project>

DummyRowDataSource.java 源数据生成样例

package com.bytedance.bytehouse.ce.examples;

import com.bytedance.bytehouse.flink.table.api.RowDataConstructor;

// CHECKSTYLE:OFF: checkstyle:AvoidStarImport
import java.util.*;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicLong;

import org.apache.flink.configuration.Configuration;
import org.apache.flink.streaming.api.functions.source.RichParallelSourceFunction;
import org.apache.flink.table.api.DataTypes;
import org.apache.flink.table.data.RowData;
import org.apache.flink.table.types.DataType;

/**
  * Synthetic {@code RowData} generator mimicking the feed of crime cases reported by Neighbourhood
  * Police Centres (NPCs) in Singapore.
  */
public class DummyRowDataSource extends RichParallelSourceFunction<RowData> {

  private static final AtomicLong pullCount = new AtomicLong();

  static Map<String, Timer> map  = new ConcurrentHashMap<>(2, 0.9f, 1);

  private final List<String> offences = Arrays.asList("Unlicensed Moneylending", "Harassment");

  private final AtomicInteger caseNo = new AtomicInteger();

  private volatile boolean cancelled = false;

  private Random random;

  @Override
  public void open(Configuration parameters) throws Exception {
    super.open(parameters);
    random = new Random();

    map.computeIfAbsent(
        "holder",
        s -> {
          final Timer timer = new Timer("RandomStringSource ", true);
          timer.schedule(
              new TimerTask() {
                @Override
                public void run() {
                  System.out.printf("source is pulled %s times\n", pullCount.get());
                }
              },
              5000,
              5000);
          return timer;
        });
  }

  private final RowDataConstructor rowDataConstructor =
      RowDataConstructor.apply(
          new DataType[] {
            DataTypes.INT(), DataTypes.STRING(), DataTypes.STRING(), DataTypes.INT()
          });

  @Override
  public void run(SourceContext<RowData> ctx) throws Exception {
    while (!cancelled) {
      Thread.sleep(random.nextInt(10) + 5);
      synchronized (ctx.getCheckpointLock()) {
        final Object[] rowDataFields = {
          2000 + random.nextInt(20),
          generateRandomWord(4),
          randomOffences(),
          caseNo.incrementAndGet()
        };
        ctx.collect(rowDataConstructor.constructInsert(rowDataFields));
        pullCount.incrementAndGet();
      }
    }
  }

  @Override
  public void cancel() {
    cancelled = true;
  }

  private String generateRandomWord(int wordLength) {
    StringBuilder sb = new StringBuilder(wordLength);
    for (int i = 0; i < wordLength; i++) { // For each letter in the word
      char tmp = (char) ('a' + random.nextInt('z' - 'a')); // Generate a letter between a and z
      sb.append(tmp); // Add it to the String
    }
    return sb.toString();
  }

  private String randomOffences() {
    return offences.get(random.nextInt(2));
  }
}

CeGatewaySinkDataStreamExample.java 数据 sink 样例

package com.bytedance.bytehouse.ce.examples;

import com.bytedance.bytehouse.flink.connector.clickhouse.ClickHouseSinkFunction;
import com.bytedance.bytehouse.flink.connector.clickhouse.api.java.ClickHouseSinkFunctionBuilder;
import java.time.Duration;
import java.util.Arrays;
import java.util.List;
import org.apache.flink.api.common.typeinfo.TypeInformation;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.table.api.DataTypes;
import org.apache.flink.table.catalog.Column;
import org.apache.flink.table.data.RowData;

// CHECKSTYLE:OFF: checkstyle:InvalidTarget
/**
  * Showcasing the usage of ByteHouse CE connector to sink data into a ByteHouse CE table.
  *
  * <p>An example of a SQL statement to create a local table compatible with this program is as
  * follows:
  *
  * <pre>
  * CREATE TABLE `demo_db`.`demo_table_local` ON CLUSTER `demo_cluster` (
  *   `year` Int32,
  *   `npc` String,
  *   `offence` String,
  *   `case_no` Int32
  * )
  * ENGINE = HaUniqueMergeTree(('/clickhouse/demo_cluster/demo_table/{shard}', '{replica}'))
  * ORDER BY `case_no`;
  * </pre>
  */
public class CeSinkDataStreamExample {

  public static void main(String[] args) throws Exception {

    final String host = args[0];
    final int port = Integer.parseInt(args[1]);
    final String username = args[2];
    final String password = args[3];
    final String clusterName = args[4];
    final String dbName = args[5];
    final String tableName = args[6];

    // Creating the execution environment
    final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

    // Adding a source to the data stream
    DataStream<RowData> dataStream =
        env.addSource(new DummyRowDataSource()).returns(TypeInformation.of(RowData.class));

    // List of columns representing the table schema
    List<Column> columns =
        Arrays.asList(
            Column.physical("year", DataTypes.INT()),
            Column.physical("npc", DataTypes.STRING()),
            Column.physical("offence", DataTypes.STRING()),
            Column.physical("case_no", DataTypes.INT()));

    try (@SuppressWarnings("unchecked")
  ClickHouseSinkFunction<RowData, ?> sink =
    new ClickHouseSinkFunctionBuilder.Upsert(clusterName, dbName, tableName)
        .withSchema(columns)
        .withShardDiscoveryKind("CE_GATEWAY")
         .withGateway(host, port)
        .withAccount(username, password)
        .withShardingKey("year")
        .withFlushInterval(Duration.ofSeconds(1))
        .build()) {

      // Add the sink to the data stream
      dataStream.addSink(sink);

      // Trigger the execution
      env.execute(CeSinkDataStreamExample.class.getSimpleName());
    }
  }
}

编译项目生成可运行 jar

为了减少潜在的包冲突文件，建议用户打 fat jar，集成相关依赖。
在项目 pom.xml 文件中，添加如下build方式，以生成 fat jar。

<build>
  <finalName>fat-jar-example</finalName>
  <plugins>
    <plugin>
      <groupId>org.apache.maven.plugins</groupId>
      <artifactId>maven-assembly-plugin</artifactId>
      <version>3.4.2</version>
      <configuration>
        <descriptorRefs>
          <descriptorRef>jar-with-dependencies</descriptorRef>
        </descriptorRefs>
      </configuration>
      <executions>
        <execution>
          <id>assemble-all</id>
          <phase>package</phase>
          <goals>
            <goal>single</goal>
          </goals>
        </execution>
      </executions>
    </plugin>
  </plugins>
</build>

上传 jar 包到 EMR 集群

可通过 scp 指令。

scp xxxx.jar root@xxx.xxx.xxx.xxx:/{path}/

执行任务

提交任务的命令：

/usr/lib/emr/current/flink/bin/flink run-application -t yarn-application <JAR 包文件名>

Yarn ResourceManager UI 查看运行状态

Yarn ResourceManager UI 访问方式，可参考访问链接 E-MapReduce-火山引擎。