使用YCSB对Clickhouse进行压测

一、前置条件

成功安装jdk及maven。

二、实现Clickhouse压测客户端

目前YCSB没有实现Clickhouse客户端，因此需要我们自己实现。
添加新database客户端的wiki：Adding a Database

拉取项目

git clone https://github.com/brianfrankcooper/YCSB.git

或者直接下载压缩包解压。

添加新module

项目根目录右键添加module。

ArtifactId为待测试客户端名称，这里为clickhouse-binding。

模块创建完成后，可以在根项目的pom文件中看到新加模块。

Clickhouse客户端实现

添加相关依赖。

<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>4.0.0modelVersion>
  <parent>
    <groupId>site.ycsbgroupId>
    <artifactId>binding-parentartifactId>
    <version>0.18.0-SNAPSHOTversion>
    <relativePath>../binding-parentrelativePath>
  parent>

  <artifactId>clickhouse-bindingartifactId>
  <name>Clickhouse Bindingname>
  <packaging>jarpackaging>

  <dependencies>
    <dependency>
      <groupId>ru.yandex.clickhousegroupId>
      <artifactId>clickhouse-jdbcartifactId>
      <version>0.1.52version>
    dependency>
    <dependency>
      <groupId>site.ycsbgroupId>
      <artifactId>coreartifactId>
      <version>${project.version}version>
      <scope>providedscope>
    dependency>
    <dependency>
      <groupId>org.slf4jgroupId>
      <artifactId>slf4j-apiartifactId>
      <version>1.7.13version>
    dependency>
    <dependency>
      <groupId>org.slf4jgroupId>
      <artifactId>slf4j-simpleartifactId>
      <version>1.7.13version>
    dependency>
    <dependency>
      <groupId>junitgroupId>
      <artifactId>junitartifactId>
      <version>4.12version>
      <scope>testscope>
    dependency>
  dependencies>

project>

客户端实现，由于这里我只需要测试查询性能，且在现有数据集上测试，因此只实现了scan()方法。另外，我测试的sql只有一个返回值，因此只用了一个key表示，若需多个返回值，可以通过传入字段数组的方式对其进行修改。

package site.ycsb.db.clickhouse;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import ru.yandex.clickhouse.ClickHouseConnection;
import ru.yandex.clickhouse.ClickHouseDataSource;
import ru.yandex.clickhouse.settings.ClickHouseProperties;
import site.ycsb.ByteIterator;
import site.ycsb.DBException;
import site.ycsb.Status;
import site.ycsb.StringByteIterator;

import java.sql.*;
import java.util.*;
import java.util.concurrent.atomic.AtomicInteger;

/**
 * Clickhouse client for YCSB framework.
 */
public class ClickhouseClient extends site.ycsb.DB {
  private static final Logger LOGGER = LoggerFactory.getLogger(ClickhouseClient.class);

  /** Count the number of times initialized to teardown on the last. */
  private static final AtomicInteger INIT_COUNT = new AtomicInteger(0);

  private static ClickHouseConnection connection;

  // clickhouse url
  private static final String CONNECTION_URL = "clickhouse.url";

  // clickhouse port
  private static final String CONNECTION_PORT = "clickhouse.port";

  // clickhouse database name
  private static final String CONNECTION_DB = "clickhouse.db";

  // clickhouse username
  private static final String CONNECTION_USER = "clickhouse.username";

  // clickhouse password
  private static final String CONNECTION_PASSWORD = "clickhouse.password";

  // sql for test
  private static final String SCAN_SQL = "clickhouse.scan_sql";

  // sql result
  private static final String RES_KEY = "clickhouse.key";

  private static final String DEFAULT_PROP = "";

  private String sql;

  private String key;

  private PreparedStatement statement;

  @Override
  public void init() throws DBException {
    INIT_COUNT.incrementAndGet();
    synchronized (ClickhouseClient.class) {
      Properties props = getProperties();
      String url = props.getProperty(CONNECTION_URL, DEFAULT_PROP);
      String port = props.getProperty(CONNECTION_PORT, DEFAULT_PROP);
      String db = props.getProperty(CONNECTION_DB, DEFAULT_PROP);
      String username = props.getProperty(CONNECTION_USER, DEFAULT_PROP);
      String password = props.getProperty(CONNECTION_PASSWORD, DEFAULT_PROP);
      sql = props.getProperty(SCAN_SQL, DEFAULT_PROP);
      key = props.getProperty(RES_KEY, DEFAULT_PROP);

      ClickHouseProperties properties = new ClickHouseProperties();
      properties.setUser(username);
      properties.setPassword(password);
      properties.setDatabase(db);
      properties.setSocketTimeout(60000000);
      ClickHouseDataSource clickHouseDataSource = new ClickHouseDataSource("jdbc:clickhouse://" + url + ":" + port, properties);
      try {
        connection = clickHouseDataSource.getConnection();
        System.out.println(connection);
        statement = connection.prepareStatement(sql);
      } catch (SQLException e) {
        LOGGER.error(e.getMessage());
      }
    }
  }

  @Override
  public void cleanup() throws DBException {
    if (INIT_COUNT.decrementAndGet() == 0) {
      try {
        if (!connection.getAutoCommit()){
          connection.commit();
        }
        connection.close();
      } catch (SQLException e) {
        System.err.println("Error in cleanup execution. " + e);
      }
    }
  }

  @Override
  public Status read(String table, String key, Set<String> fields, Map<String, ByteIterator> result) {
    return null;
  }

  @Override
  public Status scan(String table, String startkey, int recordcount, Set<String> fields, Vector<HashMap<String, ByteIterator>> result) {
    try {
      ResultSet resultSet = statement.executeQuery();
      if (!resultSet.next()) {
        resultSet.close();
        return  Status.NOT_FOUND;
      }
      HashMap<String, ByteIterator> values = new HashMap<String, ByteIterator>();
      Integer value = resultSet.getInt(key);
      values.put(key, new StringByteIterator(value.toString()));
      result.add(values);
      resultSet.close();
    } catch (SQLException e) {
      LOGGER.error(e.getMessage());
      return Status.ERROR;
    }
    return Status.OK;
  }

  @Override
  public Status update(String table, String key, Map<String, ByteIterator> values) {
    return null;
  }

  @Override
  public Status insert(String table, String key, Map<String, ByteIterator> values) {
    return null;
  }

  @Override
  public Status delete(String table, String key) {
    return null;
  }
}

实现单测。

package site.ycsb.db.clickhouse;

import org.junit.BeforeClass;
import org.junit.Test;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import ru.yandex.clickhouse.ClickHouseConnection;
import ru.yandex.clickhouse.ClickHouseDataSource;
import ru.yandex.clickhouse.settings.ClickHouseProperties;
import site.ycsb.ByteIterator;
import site.ycsb.DBException;

import java.io.IOException;
import java.net.Socket;
import java.sql.SQLException;
import java.util.HashMap;
import java.util.Properties;
import java.util.Vector;

import static org.hamcrest.CoreMatchers.not;
import static org.junit.Assert.assertThat;
import static org.junit.Assume.assumeNoException;

public class ClickhouseClientTest {
  private static final Logger LOGGER = LoggerFactory.getLogger(ClickhouseClientTest.class);

  private static ClickhouseClient clickhouseClient;

  private static ClickHouseConnection connection;

  private static final String CONNECTION_URL = "clickhouse.url";
  // your clickhouse ip
  private static final String CONNECTION_URL_INPUT = "your clickhouse ip";

  private static final String CONNECTION_PORT = "clickhouse.port";
  // your clickhouse port
  private static final String CONNECTION_PORT_INPUT = "your clickhouse port";

  private static final String CONNECTION_DB = "clickhouse.db";
  // your database name
  private static final String CONNECTION_DB_INPUT = "database name";

  private static final String CONNECTION_USER = "clickhouse.username";
  // your username
  private static final String CONNECTION_USER_INPUT = "your username";

  private static final String CONNECTION_PASSWORD = "clickhouse.password";
  // your password
  private static final String CONNECTION_PASSWORD_INPUT = "your password";

  private static final String SCAN_SQL = "clickhouse.scan_sql";
  // your test sql
  private static final String SCAN_SQL_INPUT = "select uniqExact(user_id) as count from tb";

  private static final String RES_KEY = "clickhouse.key";
  // the field returned by sql
  private static final String RES_KEY_INPUT = "count";

  @BeforeClass
  public static void setUp() {
    try (Socket socket = new Socket(CONNECTION_URL_INPUT, Integer.valueOf(CONNECTION_PORT_INPUT))){
      assertThat("Socket is not bound.", socket.getLocalPort(), not(-1));
    } catch (IOException connectFailed) {
      assumeNoException("PostgreSQL is not running. Skipping tests.", connectFailed);
    }

    Properties props = new Properties();
    props.setProperty(CONNECTION_URL, CONNECTION_URL_INPUT);
    props.setProperty(CONNECTION_PORT, CONNECTION_PORT_INPUT);
    props.setProperty(CONNECTION_DB, CONNECTION_DB_INPUT);
    props.setProperty(CONNECTION_USER, CONNECTION_USER_INPUT);
    props.setProperty(CONNECTION_PASSWORD, CONNECTION_PASSWORD_INPUT);
    props.setProperty(SCAN_SQL, SCAN_SQL_INPUT);
    props.setProperty(RES_KEY, RES_KEY_INPUT);

    try{
      ClickHouseProperties properties = new ClickHouseProperties();
      properties.setUser(CONNECTION_USER_INPUT);
      properties.setPassword(CONNECTION_PASSWORD_INPUT);
      properties.setDatabase(CONNECTION_DB_INPUT);
      properties.setSocketTimeout(600000);
      ClickHouseDataSource clickHouseDataSource = new ClickHouseDataSource("jdbc:clickhouse://" + CONNECTION_URL_INPUT + ":" + CONNECTION_PORT_INPUT, properties);
      connection = clickHouseDataSource.getConnection();

      boolean tableExists = connection.getMetaData().getTables(null, null, "tb_user_action_3", null).next();

      assertThat("Table does not exist.", tableExists, not(false));

      clickhouseClient = new ClickhouseClient();
      clickhouseClient.setProperties(props);
      clickhouseClient.init();
    } catch (SQLException | DBException e){
      LOGGER.error(e.toString());
    }
  }

  @Test
  public void scanTest() {
    Vector<HashMap<String, ByteIterator>> results = new Vector<>();
    clickhouseClient.scan("", "", 1, null, results);
    LOGGER.info(results.toString());
  }
}

压测前准备

YCSB的bin目录下事先准备好了执行脚本，但为了可以正确使用它，需要做一些准备。
首先修改bin目录下bindings.properties文件，添加clickhouse标识，使脚本可以识别clickhouse，并映射到clickhouse的客户端。

如果使用python脚本ycsb，则还需修改脚本，在DATABASES字典中添加clickhouse的映射。

为了使测试方便，我还在脚本中添加了打包时跳过代码风格检查的flag。

别的脚本同理，修改mvn命令即可。

打包

如果嫌麻烦，可以直接在项目根目录下：

mvn clean
mvn install

也可以单独给模块打包，在core、binding-parent等自定义数据库客户端依赖的包目录以及clickhouse包目录下执行：

mvn clean
mvn install

实现测试配置文件

在workloads目录下复制一份配置文件模板。

cp workload_template workload_clickhouse

添加并修改文件内容。

...

# The name of the workload class to use
workload=site.ycsb.workloads.CoreWorkload

# The inputs of clickhouse
# 完善信息
clickhouse.url=
clickhouse.port=
clickhouse.db=
clickhouse.username=
clickhouse.password=
clickhouse.scan_sql=
clickhouse.key=

# There is no default setting for operationcount but it is
# required to be set.
# The number of operations to use during the run phase.
# 总操作数
operationcount=30

...

# What proportion of operations are reads
readproportion=0

# What proportion of operations are updates
updateproportion=0

# What proportion of operations are inserts
insertproportion=0

# What proportion of operations read then modify a record
readmodifywriteproportion=0

# scan操作占比100%
# What proportion of operations are scans
scanproportion=1


...

执行脚本进行测试

在项目根目录下：

./bin/ycsb run clickhouse -threads 2 -target 10 -P ./workloads/workload_clickhouse

即可看到压测结果。