question How to create impala derived tables in Support Questions

How to create impala derived tables

Khalef — Fri, 16 Sep 2022 10:27:20 GMT

I read in a presentation that derived tables will be available in v > 2.4, I am using 2.5 with quick start but it seems that the CREATE DERIVED TABLE keyword is not supported???

Re: How to create impala derived tables

Tim Armstrong — Thu, 23 Jun 2016 20:05:23 GMT

That feature has not made it in unfortunately. The documentation at http://www.cloudera.com/documentation.html is the source of truth about what features are or are not present.

Re: How to create impala derived tables

Khalef — Fri, 24 Jun 2016 00:31:03 GMT

Thank you Tim for the quick reply.

When can we expect the feature to be available? this is my best selling point of Cloudera option for the team

Re: How to create impala derived tables

Tim Armstrong — Fri, 24 Jun 2016 01:42:41 GMT

I don't think it's on the immediate roadmap, our focus recently has been on various other things (performance, Amazon EC2 support, etc)

Re: How to create impala derived tables

Khalef — Sat, 25 Jun 2016 04:46:50 GMT

Hi Tim

That is unfortunate.

However,

I am creating external tables with configured locations and that is great, because now the only thing I need to do to push new daily data is to just create the parquet file and alter the table with the new partition details.

Now the challenge is the aggregated tables that are created as CREATE TABLE FOO AS SELECT..FROM ABC, XYZ,..WHERE...;

These ones I have to recreate them at each time and there is no way to just alter them and make them aware of the new data. The queries are evaluated at creation time!

Is there any other way to create aggregated tables that are updated automatically?

Cheers Khalef.

Re: How to create impala derived tables

Tim Armstrong — Tue, 28 Jun 2016 14:53:34 GMT

The only way to do this with zero work would be to use a view. http://www.cloudera.com/documentation/enterprise/latest/topics/impala_create_view.html

Otherwise you do have to run the queries as part of your data pipeline as you mentioned.