1、新一代超高性能的大数据集成工具 Apache SeaTunnel演讲人：代立冬(David)CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 202

2、3关关于于我我白鲸开源联合创始人Apache 基金会正式成员Apache 孵化器导师Apache SeaTunnel PMCApache DolphinScheduler PMC ChairApacheCon Asia 2022/2023 大数据论坛主席代立冬(David Zollo)CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023Core

3、JavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CONTENT目录01数据集成的痛点02SeaTunnel 功能与架构03用户案例04发展规划CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023Co

4、reJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023数据集成的痛点1CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJ

5、avaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023数据源多达几百种，版本间不兼容，而且不断有新的出现频繁读取 binlog 对数据源端影响大大事务、Schema 变更影响下游低吞吐高时延导致数据无法及时到达离线同步和实时同步常被分开管理，维护困难数据割接人工进行数据丢失与重复，无法一致性出现问题无法回滚或者断点继续执行同步过程不透明，缺少监控企业数据集成面临的问题CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWee

6、k 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 20236Apache SeaTunnel 简介2023 年 6 月 1 日正式成为 Apache 顶级项目。首个国人主导的数据集成项目每天可以稳定高效同步万亿级数据，已在数百家公司生产上使用Next-generation high-performance

7、,distributed,massive data integration toolCoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023Sea

8、Tunnel 功能与架构2CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023无中心化设计确保系统的高可用，支持多云支持每日万亿级数据量同步简

9、单易用，开箱即用，不依赖 HDFS，Flink，Spark全可视化操作存算分离架构设计高性能数据同步支持节点动态伸缩全量到增量无锁化自动切换读缓冲(一个源到多个目标数据源，只用一次读取)动态速率控制，对源端和目标端压力可控支持 Schema Evolution断点续传实现 Exactly-Once 一次语义，保证数据一致性云组件支持K8s支持AWS Redshift、S3,RDS,DynamoDB阿里 OSS File，TableStore等批流、实时、CDC一体化配置无主键增量数据集成整库同步、表结构自动变更丰富的数据源支持，目前已经支持 100+种数据源SeaTunnel 核心功能与目标C

10、oreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 20239 SeaTunnel：新一代实时多源数据同步工具.上上百百种种数数据据源源数数据据同同步步

11、与与集集成成目目标标数数据据源源原原有有解解决决方方案案Sea TunnelSeaTunnelUniversal APISeaTunnelEngine 批量数据全量、增量集成 实时数据集成 批量无主键增量集成等Table APISource APIEngine APISink API100+接接口口数数量量3x版版本本迭迭代代500+%接接口口数数增增长长MySQLPostgreSQLKafkaMongoDBElasticTiDBDruidRedisHiveHudiKuduHBaseMySQLMongoDBOracleHudiInfluxDBPostgreSQLElasticRedisDori

12、sNeo4jKafkaTiDBHiveHBaseFeishu其其他他解解决决方方案案FacebookGoogle AdsHubSpotSalesforceAirtableClickhouseSeaTunnel性性能能快快30%SeaTunnel性性能能快快30倍倍DataXCoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWe

13、ek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 202310 SeaTunnel 架构SourcesTargetsCoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaW

14、eek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023Apache SeaTunnel 设计核心理念SeaTunnel 设计的核心是将数据处理的各种行为抽象成 Plugin，主要概括为以下两点：1.上层不依赖底层，两者都依赖抽象2.流程代码与业务逻辑应该分离对于整个数据处理过程，大致可以分为以下几个流程：输输入入-转转换换-输输出出，对于更复杂的数据处理，实质上也是这几种行为的组合：sourcetransformsinkEngine independent Connector APIConnect

15、or TranslationSource ConnectorTransform Connector多多引引擎擎支支持持，Spark/Flink/ZetaSink Connector0CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWee

16、k 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023SeaTunnel 运行流程CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJava

17、Week 2023CoreJavaWeek 2023Apache SeaTunnel 架构特性多版本、多引擎支持支持多个版本的Flink引擎，完美支持Flink的Checkpoint流程Flink支持Spark微批处理模式，支持聚合提交特性Spark专为数据同步场景设计的引擎，还在开发中。SeaTunnel内部引擎，为那些没有大数据生态的企业或追求数据同步最佳体验的用户提供可选方案SeaTunnel ZetaCoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaW

18、eek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023流流批批一一体体统一了流和批的处理 API，新的 Connector 只需要按 API 实现一次，即可同时支持流处理和批处理下的数据集成。03JDBC多多复复用用/数数据据库库日日志志多多表表解解析析支持多表或整库同步，解决 JDBC 连接过多的问题；支持多表或整库数据库日志读

19、取解析，解决 CDC 多表同步场景下需要重复解析日志问题。04与与引引擎擎解解藕藕，支支持持 Flink、Spark、Zeta作作为为运运行行时时，专专为为数数据据集集成成场场景景设设计计.多多引引擎擎支支持持定义一套 SeaTunnel自己的 API，不依赖具体的执行引擎，实现一套代码可在不同的引擎上执行。01多多版版本本支支持持通过 Translation 层将 Connector 与引擎解藕，解决以往为了支持底层引擎一个新的版本，大部分Connector都需要修改代码的问题。02与具体执行引擎解藕的架构设计SeaTunnel Source APISeaTunnel Transform A

20、PISeaTunnel Sink APISeaTunnel Checkpoint APISeaTunnel Translation APICoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWe

21、ek 2023CoreJavaWeek 2023SeaTunnelConnectorSparkTranslationFlinkTranslationSparkConnectorFlinkConnectorImplementing based on SeaTunnel Connector API基基于于 Spark Connector API 将将 SeaTunnel API实实现现的的 Connector 包包装装成成 Spark Connector基基于于 Flink Connector API 将将 SeaTunnel API实实现现的的 Connector 包包装装成成 Flink Co

22、nnector运运行行在在 Spark 上上面面的的 Connector 内内部部实实现现从从 SeaTunnel Row 转转换换成成 Spark 数数据据结结构构接接口口运运行行在在 Flink上上面面的的 Connector 内内部部实实现现从从 SeaTunnel Row 转转换换成成 Flink 数数据据结结构构接接口口我们为什么能支持 Spark,Flink,or Zeta 作为执行引擎?SeaTunnel Translation LayerCoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023Cor

23、eJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 202316 SeaTunnel Zeta Engine 解决 Flink/Spark 同步集成引擎痛点问题回回滚滚容容错错容容错错粒粒度度大大进行多表同步时，Flink/Spark任何表出现问题都会导致整个作业失败停止，导致所有表同步延迟

24、多多方方式式确确保保一一致致性性支持无中心HA和更细粒度的作业回滚机制，结合多阶段提交与CheckPoint机制，确保数据一致的同时避免大量回滚导致性能下降S Sp pa ar rk k /F Fl li in nk kSeaTunnel Zeta引引擎擎资资源源把把控控资资源源浪浪费费严严重重每个作业只能同步一张表，多张表同步需要启动多个Job运行，造成巨大浪费资源易易用用省省资资源源Zeta引擎的Dynamic Thread Sharing技术可提高CPU利用率，不依赖HDFS，Spark等复杂组件，具备更好单机处理性能多多表表同同步步JDBC连连接接数数过过多多每个task只能处理一张表

25、，每张表至少需要一个JDBC连接来读取或写入数据。当进行多表同步和整库同步时，需要大量的JDBC连接极极致致CDC&批批量量性性能能支持多表或整库同步，解决JDBC连接过多的问题；实现 zero-copy 技术，无需序列化开销，列式内存格式增加大吞吐CDC场场景景不不支支持持数数据据缓缓存存与与表表结结构构变变更更CDC场景下易出现源端数据库日志被清除的情况，同时目前Datax/Spark/Flink/FlinkCDC都无法支持DDL变更的检测和下游应用非常重要多多复复用用/数数据据库库日日志志多多表表解解析析支持多表或整库数据库日志读取解析，解决CDC多表同步场景下需要重复解析日志的问题灵灵

26、活活复复用用开开发发与与运运维维困困难难需要投入的人力完成安装、运维，保证正常运转;同时需要投入较多开发人员编程实现各种复杂的业务计算快快速速开开发发与与运运维维SeaTunnel 帮助用户快速建立单表多字段、多表多字段、SaaS和非结构化到数据库等复杂的数据集成任务并监控CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWe

27、ek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023SeaTunnel Zeta 引擎不需要依赖三方组件，不依赖大数据平台无主（自选主）WAL，整个集群重启也可恢复之前正在运行的作业支持分布式快照算法，保障数据一致性CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJav

28、aWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023SeaTunnel Zeta 性能对比本地测试场景：MySQL-Hive,Postgres-Hive,SQLServer-Hive,Orache-Hive云测试场景：MySQL-S3列数：32，基本包含大部分数据类型行数：3000w 行Hive 文件 text 格式 18G测试节点：单机 8C16G本地测试：SeaTunnel Zeta V

29、S DataXSeaTunnel Zeta 比 DataX 同步数据快 30-50%左右内存对 SeaTunnel Zeta 的性能没有影响云数据同步：SeaTunnel 在 MySQL 到 S3 场景下性能是 Airbyte 的 30 多倍，是 AWS DMS 和 Glue 的 2 到 5 倍CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 202

30、3CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023SeaTunnel 现状-数据源量支持的数据源有 100+种CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2

31、023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023Apache SeaTunnel vs 相关项目对对比比项项Apache SeaTunnelDataXApache SqoopApache Flume部署难度容易容易十分复杂，严重依赖 Hadoop 体系容易运行模式分布式，也支持单机单机本身不是分布式框架，依赖 Hadoop MR 实现分布式分布式，也支持单机健壮的容错机制无中心化的高可用架构设计，有完善的容错机制易受比如网络闪断、数据源不稳定等因素影响MR 模式重，出

32、错处理麻烦一般支持的数据源丰富度支持 MySQL、PostgreSQL、Oracle、SQLServer、S3、RedShift、HBase、Clickhouse、Hive等过 100 种数据源支持 MySQL、ODPS、PostgreSQL、Oracle、Hive 等 20+种数据源仅支持 MySQL、Oracle、DB2、Hive、HBase、S3 等几种数据源支持 Kafka、File、HTTP、Avro、HDFS、Hive、HBase等几种数据源自动建表支持不支持不支持不支持整库同步支持不支持不支持不支持断点续传支持不支持不支持不支持多引擎支持支持 SeaTunnel Zeta、Fli

33、nk、Spark 3 个引擎选其一作为运行时只能跑在 DataX 自己引擎上自身无引擎，需跑在 Hadoop MR 上，任务启动速度非常慢支持 Flume 自身引擎数据转换(Transform)支持 Copy、Filter、Replace、Split、SQL、自定义 UDF 等算子支持补全，过滤等算子只有列映射、数据类型转换和数据过滤基本算子只支持 Interceptor 方式简单转换操作单机性能比 DataX 高 20%-50%较好一般一般离线同步支持支持支持支持增量同步支持支持支持支持实时同步支持不支持不支持支持CDC同步支持不支持不支持不支持批流一体支持不支持不支持不支持精确一致性MyS

34、QL、Kafka、Hive、HDFS、File 等连接器支持不支持不支持不支持，提供一定程度的一致性可扩展性插件机制非常易扩展易扩展扩展性有限，Sqoop主要用于将数据在Apache Hadoop和关系型数据库之间传输易扩展统计信息有有无有Web UI正在实现中(拖拉拽即可完成)无无无与调度系统集成度已经与 DolphinScheduler 集成，后续也会支持其他调度系统不支持 不支持不支持社区非常活跃非常不活跃已经从 Apache 退役不活跃CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJava

35、Week 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023./bin/seatunnel.sh-config./config/v2.batch.config.template-e local3 分钟入门案例参考：https:/ job.mode=“BATCH”source FakeSource pa

36、rallelism=2 result_table_name=“fake”row.num=16 schema=fields name=“string”age=“int”sink Console Fake-ConsoleSo easy!SeaTunnel 使用方式非常简单CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek

37、2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023SeaTunnel 使用方式非常简单 MySQL-Doris#定义一些作业的运行参数,具体可以参考 https:/seatunnel.apache.org/docs/2.3.1/concept/JobEnvConfigenv job.mode=“BATCH”#作业的运行模式，BATCH=离线批同步，STREAMING=实时同步 job.name=“SeaTunnel_Job”checkpoint.interval=10

38、000#每10000ms进行一次checkpoint，后面会详细介绍checkpoint对JDBC Source和StarRocks Sink这两个连接器的影响source Jdbc parallelism=5#并行度，这里是启动5个Source Task来并行的读取数据 partition_column=“id”#使用id字段来进行split的拆分，目前只支持数字类型的主键列，而且该列的值最好是离线的，自增id最佳 partition_num=“20”#拆分成20个split，这20个split会被分配给5个Source Task来处理 result_table_name=“Table921

39、0050164000”query=“SELECT id,f_binary,f_blob,f_long_varbinary,f_longblob,f_tinyblob,f_varbinary,f_smallint,f_smallint_unsigned,f_mediumint,f_mediumint_unsigned,f_int,f_int_unsigned,f_integer,f_integer_unsigned FROM sr_test.test1”password=“root123”driver=“com.mysql.cj.jdbc.Driver”user=root url=“jdbc:m

40、ysql:/st01:3306/sr_test?enabledTLSProtocols=TLSv1.2&rewriteBatchedStatements=true”sink Doris fenodes=e2e_dorisdb:8030 username=root password=table.identifier=test.e2e_table_sink sink.enable-2pc=true sink.label-prefix=test_csv doris.config=format=csv column_separator=,参考：https:/ 2023CoreJavaWeek 2023

41、CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023用户案例3CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJa

42、vaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 202324 SeaTunnel 典典型型案案例例多多源源数数据据高高频频出出入入数数据据仓仓库库异异构构数数据据实实时时数数据据同同步步解决多数据源数据每日出入数据库以及每日出入仓同步数据问题，数据集群规模几十台，日均记录数上上千千亿亿，日均数

43、据量在 100T 以上。解决从 MySql，日志文件、Presto、Kafka、Spark、ClickHouse 以及 Hudi 之间数据同步问题，覆盖数十台集群。CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2

44、023CoreJavaWeek 2023CoreJavaWeek 202325SeaTunnel 在哔哩哔哩的落地实践CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023Cor

45、eJavaWeek 2023SeaTunnel 发展规划4CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023SeaTunnel 发展历程项目

46、发展历程与规划20172021.112021.122022.032022.102022.11 首首个个 Apache 版版本本发发布布进进入入Apache孵孵化化器器并并更更名名为为SeaTunnel发发布布版版本本 30+腾腾讯讯、新新浪浪、等等上上百百家家企企业业生生产产使使用用开源 Waterdrop新一代数据同步引擎Zeta 发布发发布布第第一一个个重重大大版版本本 2.2.0，实实现现跨跨引引擎擎的的连连接接器器支支持持支支持持CDC同同步步，连连接接器器个个数数突突破破100+2022.122023-5支支持持Flink15/Spark3Zeta引引擎擎支支持持CDC整整库库同同步

47、步和和多多表表同同步步支支持持自自动动建建表表CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023更更快快、更更好好用用作为一个数据集成平台，

48、SeaTunnel将不断专注于解决数据集成领域的需求和问题。持续从数据源的数量、数据同步的性能和易用性上满足用户的需求。连连接接器器丰丰富富数据源向量数据库发发布布 SeaTunnel Web可视化作业管理编程式和引导式作业配置。内部调度+三方调度CDC 支支持持 DDL 变变更更流流速速控控制制SeaTunnel RoadmapK8S 支支持持CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJav

49、aWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023欢迎参与 SeaTunnel 社区贡献 寻找你感兴趣的 issuehttps:/ 参考贡献指南https:/ 连接器极简开发流程】https:/ API Connector 开发解析】https:/ 与 Sink API 设计解析】参与讨论&寻求帮助在邮件列表、Slack 中讨论通过微信群沟通（如果没有加入请关注 SeaTunnel 公众号

50、入群）参与 PR Review 发表你的见解CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 202330 SeaTunnel 相关资源官网:htt

51、ps:/seatunnel.apache.orgGitHub：https:/ 站：https:/ 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023感 谢 聆 听演讲人：嘉宾名CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023CoreJavaWeek 2023