[
https://issues.apache.org/jira/browse/HUDI-388?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vinoth Chandar updated HUDI-388:
--------------------------------
Status: New (was: Open)
> Support DDL / DML SparkSQL statements which useful for admins
> -------------------------------------------------------------
>
> Key: HUDI-388
> URL: https://issues.apache.org/jira/browse/HUDI-388
> Project: Apache Hudi (incubating)
> Issue Type: New Feature
> Components: CLI
> Reporter: lamber-ken
> Assignee: lamber-ken
> Priority: Major
>
> *Purpose*
> Currently, hudi offers some tools available to operate an ecosystem of Hudi
> datasets including hudi-cli, metrics, spark ui[1]. It's easy for admins to
> manage the hudi datasets by some customized ddl sql statements instead of via
> hudi-cli.
>
> After SPARK-18127, we can customize the spark session with our own optimizer,
> parser, analyzer, and physical plan strategy rules in Spark. Here are some
> steps to extend spark session
> 1, Need a tool to parse the SparkSQL statements, like antlr, RegExp.
> 2, A class which extends org.apache.spark.sql.SparkSessionExtensions and
> inject the parser.
> 3, Run the customized statements by extending
> org.apache.spark.sql.execution.command.RunnableCommand.
>
> *Demo*
> 1, Extend SparkSessionExtensions
> {code:java}
> class HudiSparkSessionExtension extends (SparkSessionExtensions => Unit) {
> override def apply(extensions: SparkSessionExtensions): Unit = {
> extensions.injectParser { (session, parser) =>
> new HudiDDLParser(parser)
> }
> }
> } {code}
>
> 2, Extend RunnableCommand
> {code:java}
> case class HudiStatCommand(path: String) extends RunnableCommand {
> override val output: Seq[Attribute] = {
> Seq(
> AttributeReference("CommitTime", StringType, nullable = false)(),
> AttributeReference("Total Upserted", IntegerType, nullable = false)(),
> AttributeReference("Total Written", IntegerType, nullable = false)(),
> AttributeReference("Write Amplifiation Factor", DoubleType, nullable =
> false)()
> )
> }
> override def run(sparkSession: SparkSession): Seq[Row] = {
> Seq(
> Row("20191207003131", 0, 10, 0.1),
> Row("20191207003200", 4, 10, 2.50),
> Row("Total", 4, 20, 5.00)
> )
> }
> }
> {code}
>
> 3, demo result, mock data
> {code:java}
> +--------------+--------------+-------------+-------------------------+
> | CommitTime|Total Upserted|Total Written|Write Amplifiation Factor|
> +--------------+--------------+-------------+-------------------------+
> |20191207003131| 0| 10| 0.1|
> |20191207003200| 4| 10| 2.5|
> | Total| 4| 20| 5.0|
> +--------------+--------------+-------------+-------------------------+
> {code}
>
> [https://github.com/lamber-ken/hudi-work]
> [http://hudi.apache.org/admin_guide.html]
> https://issues.apache.org/jira/browse/SPARK-18127
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
