English | 中文

Introduction

Linkis builds a layer of computation middleware between upper applications and underlying engines. By using standard interfaces such as REST/WS/JDBC provided by Linkis, the upper applications can easily access the underlying engines such as MySQL/Spark/Hive/Presto/Flink, etc., and achieve the intercommunication of user resources like unified variables, scripts, UDFs, functions and resource files at the same time.

As a computation middleware, Linkis provides powerful connectivity, reuse, orchestration, expansion, and governance capabilities. By decoupling the application layer and the engine layer, it simplifies the complex network call relationship, and thus reduces the overall complexity and saves the development and maintenance costs as well.

Since the first release of Linkis in 2019, it has accumulated more than 700 trial companies and 1000+ sandbox trial users, which involving diverse industries, from finance, banking, tele-communication, to manufactory, internet companies and so on. Lots of companies have already used Linkis as a unified entrance for the underlying computation and storage engines of the big data platform.

linkis-intro-01

linkis-intro-03

Features

  • Support for diverse underlying computation storage engines : Spark, Hive, Python, Shell, Flink, JDBC, Pipeline, Sqoop, OpenLooKeng, Presto, ElasticSearch, Trino, SeaTunnel, etc.;

  • Support for diverse language : SparkSQL, HiveSQL, Python, Shell, Pyspark, Scala, JSON and Java;

  • Powerful computing governance capability : It can provide task routing, load balancing, multi-tenant, traffic control, resource control and other capabilities based on multi-level labels;

  • Support full stack computation/storage engine : The ability to receive, execute and manage tasks and requests for various compute and storage engines, including offline batch tasks, interactive query tasks, real-time streaming tasks and data lake tasks;

  • Unified context service : supports cross-user, system and computing engine to associate and manage user and system resource files (JAR, ZIP, Properties, etc.), result sets, parameter variables, functions, UDFs, etc., one setting, automatic reference everywhere;

  • Unified materials : provides system and user level material management, can share and flow, share materials across users, across systems;

  • Unified data source management : provides the ability to add, delete, check and change information of Hive, ElasticSearch, Mysql, Kafka, MongoDB and other data sources, version control, connection test, and query metadata information of corresponding data sources;

  • Error code capability : provides error codes and solutions for common errors of tasks, which is convenient for users to locate problems by themselves;

Engine Type

Engine nameSupport underlying component version
(default dependency version)
Linkis Version RequirementsIncluded in Release Package By DefaultDescription
SparkApache >= 2.0.0,
CDH >= 5.4.0,
(default Apache Spark 3.2.1)
>=1.0.3YesSpark EngineConn, supports SQL , Scala, Pyspark and R code
HiveApache >= 1.0.0,
CDH >= 5.4.0,
(default Apache Hive 3.1.3)
>=1.0.3YesHive EngineConn, supports HiveQL code
PythonPython >= 2.6,
(default Python2*)
>=1.0.3YesPython EngineConn, supports python code
ShellBash >= 2.0>=1.0.3YesShell EngineConn, supports Bash shell code
JDBCMySQL >= 5.0, Hive >=1.2.1,
(default Hive-jdbc 2.3.4)
>=1.0.3NoJDBC EngineConn, already supports ClickHouse, DB2, DM, Greenplum, kingbase, MySQL, Oracle, PostgreSQL and SQLServer, can be extended quickly Support other DB, such as SQLite
FlinkFlink >= 1.12.2,
(default Apache Flink 1.12.2)
>=1.0.2NoFlink EngineConn, supports FlinkSQL code, also supports starting a new Yarn in the form of Flink Jar Application
Pipeline->=1.0.2NoPipeline EngineConn, supports file import and export
openLooKengopenLooKeng >= 1.5.0,
(default openLookEng 1.5.0)
>=1.1.1NoopenLooKeng EngineConn, supports querying data virtualization engine with Sql openLooKeng
SqoopSqoop >= 1.4.6,
(default Apache Sqoop 1.4.6)
>=1.1.2NoSqoop EngineConn, support data migration tool Sqoop engine
PrestoPresto >= 0.180>=1.2.0NoPresto EngineConn, supports Presto SQL code
ElasticSearchElasticSearch >=6.0>=1.2.0NoElasticSearch EngineConn, supports SQL and DSL code
TrinoTrino >=371>=1.3.1NoTrino EngineConn, supports Trino SQL code
SeatunnelSeatunnel >=2.1.2>=1.3.1NoSeatunnel EngineConn, supportt Seatunnel SQL code

Download

Please go to the Linkis Releases Page to download a compiled distribution or a source code package of Linkis.

Compile and Deploy

For more detailed guidance see:


Note: If you want use `-Dlinkis.build.web=true` to build linkis-web image, you need to compile linkis-web first. ## compile backend ### Mac OS/Linux # 1. When compiling for the first time, execute the following command first ./mvnw -N install # 2. make the linkis distribution package # - Option 1: make the linkis distribution package only ./mvnw clean install -Dmaven.javadoc.skip=true -Dmaven.test.skip=true # - Option 2: make the linkis distribution package and docker image # - Option 2.1: image without mysql jdbc jars ./mvnw clean install -Pdocker -Dmaven.javadoc.skip=true -Dmaven.test.skip=true # - Option 2.2: image with mysql jdbc jars ./mvnw clean install -Pdocker -Dmaven.javadoc.skip=true -Dmaven.test.skip=true -Dlinkis.build.with.jdbc=true # - Option 3: linkis distribution package and docker image (included web) ./mvnw clean install -Pdocker -Dmaven.javadoc.skip=true -Dmaven.test.skip=true -Dlinkis.build.web=true # - Option 4: linkis distribution package and docker image (included web and ldh (hadoop all in one for test)) ./mvnw clean install -Pdocker -Dmaven.javadoc.skip=true -Dmaven.test.skip=true -Dlinkis.build.web=true -Dlinkis.build.ldh=true -Dlinkis.build.with.jdbc=true ### Windows mvnw.cmd -N install mvnw.cmd clean install -Dmaven.javadoc.skip=true -Dmaven.test.skip=true ## compile web cd linkis/linkis-web npm install npm run build

Bundled with MySQL JDBC Driver

Due to the MySQL licensing restrictions, the MySQL Java Database Connectivity (JDBC) driver is not bundled with the official released linkis image by default. However, at current stage, linkis still relies on this library to work properly. To solve this problem, we provide a script which can help to creating a custom image with mysql jdbc from the official linkis image by yourself, the image created by this tool will be tagged as linkis:with-jdbc by default.

$> LINKIS_IMAGE=linkis:1.3.1 
$> ./linkis-dist/docker/scripts/make-linkis-image-with-mysql-jdbc.sh

Please refer to Quick Deployment to do the deployment.

Examples and Guidance

Documentation & Vedio

Architecture

Linkis services could be divided into three categories: computation governance services, public enhancement services and microservice governance services

  • The computation governance services, support the 3 major stages of processing a task/request: submission -> preparation -> execution
  • The public enhancement services, including the material library service, context service, and data source service
  • The microservice governance services, including Spring Cloud Gateway, Eureka and Open Feign

Below is the Linkis architecture diagram. You can find more detailed architecture docs in Linkis-Doc/Architecture. architecture

Contributing

Contributions are always welcomed, we need more contributors to build Linkis together. either code, or doc, or other supports that could help the community.
For code and documentation contributions, please follow the contribution guide.

Contact Us

  • Any questions or suggestions please kindly submit an issue.
  • By mail dev@linkis.apache.org
  • You can scan the QR code below to join our WeChat group to get more immediate response

Who is Using Linkis

We opened an issue [Who is Using Linkis] for users to feedback and record who is using Linkis.
Since the first release of Linkis in 2019, it has accumulated more than 700 trial companies and 1000+ sandbox trial users, which involving diverse industries, from finance, banking, tele-communication, to manufactory, internet companies and so on.