spark-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mahender Sarangam <Mahender.BigD...@outlook.com>
Subject Re: Any Dynamic Compilation of Scala Query
Date Tue, 08 Nov 2016 23:22:23 GMT
Hi Kiran,

Thanks for responding. We would like to know how industry is dealing scenario like Update
in SPARK.  Here is our scenario Manjunath, We are in process of migrating our SQL server data
to Spark. We have our logic in stored procedure, where we dynamically create SQL String and
execute that SQL String (Dynamic SQL), we would like to implement Dynamic string and submit
to hive context and execute it .

Here is the query in SQL

UPDATE     table1
      SET       X = A
        ,Y  = B
 FROM     Table1
 WHERE  ISNULL([Z] ,'') <> ''
      AND  [ColumnW] NOT IN ('X' ,'ACD', 'A', 'B', 'C')
     AND   [ColumnA] IS NULL


We would like to convert  using Spark SQL , the other way i would think of is using of Data
frame with "WithColumn" along with WHEN condition for each column i.e X and Y , here when
condition will have same repetitive code applied on each column based on above where clause
stmt/condition . I would like to know Industry practices for these kind of scenarios.

On 10/26/2016 4:09 AM, Manjunath, Kiran wrote:
Hi,

Can you elaborate with sample example on why you would want to do so?
Ideally there would be a better approach than solving such problems as mentioned below.

A sample example would help to understand the problem.

Regards,
Kiran

From: Mahender Sarangam <Mahender.BigData@outlook.com><mailto:Mahender.BigData@outlook.com>
Date: Wednesday, October 26, 2016 at 2:05 PM
To: user <user@spark.apache.org><mailto:user@spark.apache.org>
Subject: Any Dynamic Compilation of Scala Query

Hi,

Is there any way to dynamically execute a string  which has scala code
against spark engine. We are dynamically creating scala file, we would
like to submit this scala file to Spark, but currently spark accepts
only JAR file has input from Remote Job submission. Is there any other
way to submit .SCALA instead of .JAR to REST API of Spark ?

/MS



Mime
View raw message