CDC 是 Change Data Capture（变更数据获取）的简称。核心思想是，监测并捕获数据库的变动（包括数据或数据表的插入、更新以及删除等），将这些变更按发生的顺序完整记录下来，写入到消息中间件中以供其他服务进行订阅及消费。
CDC 主要分为基于查询和基于 Binlog 两种方式，我们主要了解一下这两种之间的区别：
| 基于查询的 CDC |
基于 Binlog 的 CDC |
|
| 开源产品 |
Sqoop、Kafka JDBC Source |
Canal、Maxwell、Debezium |
| 执行模式 |
Batch |
Streaming |
| 是否可以捕获所有数据变化 |
否 |
是 |
| 延迟性 |
高延迟 |
低延迟 |
| 是否增加数据库压力 |
是 |
否 |
Flink 社区开发了 flink-cdc-connectors 组件，这是一个可以直接从 MySQL、PostgreSQL等数据库直接读取全量数据和增量变更数据的 source 组件。目前也已开源，开源地址：https://github.com/ververica/flink-cdc-connectors
<dependencies>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-java</artifactId>
<version>1.12.0</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-java_2.12</artifactId>
<version>1.12.0</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-clients_2.12</artifactId>
<version>1.12.0</version>
</dependency>
<dependency>
<groupId>org.apache.hadoop</groupId>
<artifactId>hadoop-client</artifactId>
<version>3.1.3</version>
</dependency>
<dependency>
<groupId>mysql</groupId>
<artifactId>mysql-connector-java</artifactId>
<version>5.1.49</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-table-planner-blink_2.12</artifactId>
<version>1.12.0</version>
</dependency>
<dependency>
<groupId>com.ververica</groupId>
<artifactId>flink-connector-mysql-cdc</artifactId>
<version>2.0.0</version>
</dependency>
<dependency>
<groupId>com.alibaba</groupId>
<artifactId>fastjson</artifactId>
<version>1.2.75</version>
</dependency>
</dependencies>
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-assembly-plugin</artifactId>
<version>3.0.0</version>
<configuration>
<descriptorRefs>
<descriptorRef>jar-with-dependencies</descriptorRef>
</descriptorRefs>
</configuration>
<executions>
<execution>
<id>make-assembly</id>
<phase>package</phase>
<goals>
<goal>single</goal>
</goals>
</execution>
</executions>
</plugin>
</plugins>
</build>package com.atguigu;
import com.ververica.cdc.connectors.mysql.MySqlSource;
import com.ververica.cdc.connectors.mysql.table.StartupOptions;
import com.ververica.cdc.debezium.DebeziumSourceFunction;
import com.ververica.cdc.debezium.StringDebeziumDeserializationSchema;
import org.apache.flink.runtime.state.filesystem.FsStateBackend;
import org.apache.flink.streaming.api.CheckpointingMode;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
public class FlinkCDC {
public static void main(String[] args) throws Exception {
//1.获取Flink 执行环境
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(1);
//1.1 开启CK
// env.enableCheckpointing(5000);
// env.getCheckpointConfig().setCheckpointTimeout(10000);
// env.getCheckpointConfig().setCheckpointingMode(CheckpointingMode.EXACTLY_ONCE);
// env.getCheckpointConfig().setMaxConcurrentCheckpoints(1);
//
// env.setStateBackend(new FsStateBackend("hdfs://hadoop102:8020/cdc-test/ck"));
//2.通过FlinkCDC构建SourceFunction
DebeziumSourceFunction<String> sourceFunction = MySqlSource.<String>builder()
.hostname("hadoop102")
.port(3306)
.username("root")
.password("000000")
.databaseList("cdc_test")
//不配置表的话，就会监听库下所有的表的数据
// .tableList("cdc_test.user_info")
.deserializer(new StringDebeziumDeserializationSchema())
.startupOptions(StartupOptions.initial())
.build();
DataStreamSource<String> dataStreamSource = env.addSource(sourceFunction);
//3.数据打印
dataStreamSource.print();
//4.启动任务
env.execute("FlinkCDC");
}
}
StartupOptions:
initial:
第一次启动时 读取原表已有的历史数据, 操作类型为READ, 之后不断做检查点存储
第二次启动时 一定要指明检查点文件的具体位置, 这样就可以断点续传; 即使Flink宕机了, 重启后是从上次offset开始读, 而不是latest
检查点在打包部署后才有用, 因为那样才可以指明检查点的具体位置
earliest:
从BinLog第一行数据开始读, 最好先给这个数据库加上BinLog后, 再去读取创建数据库
latest:
读取最新变更数据, 从Flink程序启动后开始算
timestamp:
可以从BinLog某一时刻的数据开始读
specificOffset:
指明BinLog文件位置和从哪个offset开始读;
这个一般来说不怎么用, 因为本地没存offset的信息, 很难知道offset读到哪了
2）开启 MySQL Binlog 并重启 MySQL
3）启动 Flink 集群
[atguigu@hadoop102 flink-standalone]$ bin/start-cluster.sh
4）启动 HDFS 集群
[atguigu@hadoop102 flink-standalone]$ start-dfs.sh
5）启动程序
[atguigu@hadoop102 flink-standalone]$ bin/flink run -c com.atguigu.FlinkCDC flink-1.0-SNAPSHOT-jar-with-dependencies.jar
6）在 MySQL 的 gmall-flink.z_user_info 表中添加、修改或者删除数据
7）给当前的 Flink 程序创建 Savepoint
[atguigu@hadoop102 flink-standalone]$ bin/flink savepoint JobId hdfs://hadoop102:8020/flink/save
8）关闭程序以后从 Savepoint 重启程序
[atguigu@hadoop102 flink-standalone]$ bin/flink run -s hdfs://hadoop102:8020/flink/save/... -c com.atguigu.FlinkCDC flink-1.0-SNAPSHOT-jar-with-dependencies.jar
package com.atguigu;
import org.apache.flink.api.java.tuple.Tuple2;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.table.api.Table;
import org.apache.flink.table.api.bridge.java.StreamTableEnvironment;
import org.apache.flink.types.Row;
public class FlinkSQLCDC {
public static void main(String[] args) throws Exception {
//1.获取执行环境
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(1);
StreamTableEnvironment tableEnv = StreamTableEnvironment.create(env);
//2.使用FLINKSQL DDL模式构建CDC 表
tableEnv.executeSql("CREATE TABLE user_info ( " +
" id STRING primary key, " +
" name STRING, " +
" sex STRING " +
") WITH ( " +
" 'connector' = 'mysql-cdc', " +
" 'scan.startup.mode' = 'latest-offset', " +
" 'hostname' = 'hadoop102', " +
" 'port' = '3306', " +
" 'username' = 'root', " +
" 'password' = '000000', " +
" 'database-name' = 'cdc_test', " +
" 'table-name' = 'user_info' " +
")");
//3.查询数据并转换为流输出
Table table = tableEnv.sqlQuery("select * from user_info");
DataStream<Tuple2<Boolean, Row>> retractStream = tableEnv.toRetractStream(table, Row.class);
retractStream.print();
//4.启动
env.execute("FlinkSQLCDC");
}
}
2.3.1 代码实现
package com.atguigu.func;
import com.alibaba.fastjson.JSONObject;
import com.ververica.cdc.debezium.DebeziumDeserializationSchema;
import io.debezium.data.Envelope;
import org.apache.flink.api.common.typeinfo.BasicTypeInfo;
import org.apache.flink.api.common.typeinfo.TypeInformation;
import org.apache.flink.util.Collector;
import org.apache.kafka.connect.data.Field;
import org.apache.kafka.connect.data.Schema;
import org.apache.kafka.connect.data.Struct;
import org.apache.kafka.connect.source.SourceRecord;
import java.util.List;
public class CustomerDeserializationSchema implements DebeziumDeserializationSchema<String> {
/**
* {
* "db":"",
* "tableName":"",
* "before":{"id":"1001","name":""...},
* "after":{"id":"1001","name":""...},
* "op":""
* }
*/
@Override
public void deserialize(SourceRecord sourceRecord, Collector<String> collector) throws Exception {
//创建JSON对象用于封装结果数据
JSONObject result = new JSONObject();
//获取库名&表名
String topic = sourceRecord.topic();
String[] fields = topic.split("\\.");
result.put("db", fields[1]);
result.put("tableName", fields[2]);
//获取before数据
Struct value = (Struct) sourceRecord.value();
Struct before = value.getStruct("before");
JSONObject beforeJson = new JSONObject();
if (before != null) {
//获取列信息
Schema schema = before.schema();
List<Field> fieldList = schema.fields();
for (Field field : fieldList) {
beforeJson.put(field.name(), before.get(field));
}
}
result.put("before", beforeJson);
//获取after数据
Struct after = value.getStruct("after");
JSONObject afterJson = new JSONObject();
if (after != null) {
//获取列信息
Schema schema = after.schema();
List<Field> fieldList = schema.fields();
for (Field field : fieldList) {
afterJson.put(field.name(), after.get(field));
}
}
result.put("after", afterJson);
//获取操作类型
Envelope.Operation operation = Envelope.operationFor(sourceRecord);
result.put("op", operation);
//输出数据
collector.collect(result.toJSONString());
}
@Override
public TypeInformation<String> getProducedType() {
return BasicTypeInfo.STRING_TYPE_INFO;
}
}
本章图片来源于北京站 Flink Meetup 分享的《详解 Flink-CDC》
痛点：
在对于有主键的表做初始化模式，整体的流程主要分为 5 个阶段：
1.Chunk 切分；2.Chunk 分配；（实现并行读取数据&CheckPoint）
3.Chunk 读取；（实现无锁读取）
4.Chunk 汇报；
5.Chunk 分配。
根据 Netflix DBlog 的论文中的无锁算法原理，对于目标表按照主键进行数据分片，设置每个切片的区间为左闭右开或者左开右闭来保证数据的连续性。
将划分好的 Chunk 分发给多个 SourceReader，每个 SourceReader 读取表中的一部分数据，实现了并行读取的目标。
同时在每个 Chunk 读取的时候可以单独做 CheckPoint，某个 Chunk 读取失败只需要单独执行该 Chunk 的任务，而不需要像 1.x 中失败了只能从头读取。
若每个 SourceReader 保证了数据一致性，则全表就保证了数据一致性。
读取可以分为 5 个阶段
1）SourceReader 读取表数据之前先记录当前的 Binlog 位置信息记为低位点；
2）SourceReader 将自身区间内的数据查询出来并放置在 buffer 中；
3）查询完成之后记录当前的 Binlog 位置信息记为高位点；
4）在增量部分消费从低位点到高位点的 Binlog；
5）根据主键，对 buffer 中的数据进行修正并输出。
通过以上 5 个阶段可以保证每个 Chunk 最终的输出就是在高位点时该 Chunk 中最新的数据，
但是目前只是做到了保证单个 Chunk 中的数据一致性。
在 Snapshot Chunk 读取完成之后，有一个汇报的流程，如上图所示，即 SourceReader 需要将 Snapshot Chunk 完成信息汇报给 SourceEnumerator。
FlinkCDC 是支持全量+增量数据同步的，在 SourceEnumerator 接收到所有的 Snapshot Chunk 完成信息之后，还有一个消费增量数据（Binlog）的任务，此时是通过下发 Binlog Chunk给任意一个 SourceReader 进行单并发读取来实现的。
MySqlHybridSplitAssigner
private MySqlBinlogSplit createBinlogSplit(){
final List<MySqlSnapshotSplit> assignedSnapshotSplit=snapshotSplitAssigner.getAssignedSplits().values().stream().sorted(Comparator.comparing(MySqlSplit::splitId)).collect(Collectors.toList());
Map<String, BinlogOffset> splitFinishedOffsets=snapshotSplitAssigner.getSplitFinishedOffsets();
final List<FinishedSnapshotSplitInfo> finishedSnapshotSplitInfos=new ArrayList<>();
final Map<TableId, TableChanges.TableChange>tableSchemas=new HashMap<>();
BinlogOffset minBinlogOffset=BinlogOffset.INITIAL_OFFSET;for(MySqlSnapshotSplit split:assignedSnapshotSplit){
// find the min binlog offset
BinlogOffset binlogOffset=splitFinishedOffsets.get(split.splitId());if(binlogOffset.compareTo(minBinlogOffset)< 0){
minBinlogOffset=binlogOffset;
}
finishedSnapshotSplitInfos.add(
new FinishedSnapshotSplitInfo(
split.getTableId(),
split.splitId(),
split.getSplitStart(),
split.getSplitEnd(),
binlogOffset));
tableSchemas.putAll(split.getTableSchemas());
}
final MySqlSnapshotSplit lastSnapshotSplit=assignedSnapshotSplit.get(assignedSnapshotSplit.size()-1).asSnapshotSplit();
return new MySqlBinlogSplit(
BINLOG_SPLIT_ID,
lastSnapshotSplit.getSplitKeyType(),
minBinlogOffset,
BinlogOffset.NO_STOPPING_OFFSET,
finishedSnapshotSplitInfos,
tableSchemas);
}3.4.2 读取低位点到高位点之间的 Binlog
BinlogSplitReader
/**
*Returns the record should emit or not.
*
*<p>The watermark signal algorithm is the binlog split reader only sends the binlog event
that
*belongs to its finished snapshot splits. For each snapshot split, the binlog event is valid
*since the offset is after its high watermark.
*
* <pre> E.g: the data input is :
* snapshot-split-0 info : [0, 1024) highWatermark0
*snapshot-split-1 info : [1024, 2048) highWatermark1
*the data output is:
*only the binlog event belong to [0, 1024) and offset is after highWatermark0 should send,
*only the binlog event belong to [1024, 2048) and offset is after highWatermark1 should
send.
*</pre>
*/
private boolean shouldEmit(SourceRecord sourceRecord) {
if (isDataChangeRecord(sourceRecord)) {
TableId tableId = getTableId(sourceRecord);
BinlogOffset position = getBinlogPosition(sourceRecord);
//aligned, all snapshot splits of the table has reached max highWatermark if (position.isAtOrBefore(maxSplitHighWatermarkMap.get(tableId))) {
return true;
}
Object[] key = getSplitKey(
currentBinlogSplit.getSplitKeyType(),
sourceRecord,
statefulTaskContext.getSchemaNameAdjuster()); for (FinishedSnapshotSplitInfo splitInfo : finishedSplitsInfo.get(tableId)) {
if (RecordUtils.splitKeyRangeContains(
key, splitInfo.getSplitStart(), splitInfo.getSplitEnd())
&&position.isAtOrBefore(splitInfo.getHighWatermark())) { return true;
}
}
//not in the monitored splits scope, do not emit return false;
}
//always send the schema change event and signal event
//we need record them to state of Flink
return true;
}