FAQ
If I have a set of objects of Transaction packed into Parquet format, how
can I sum the total sales of all transactions in Impala?

class LineItem {
   val unit: Int
   val price: Double
}

class Transaction {
    val lineItems: Array[LineItem]
}

will this query work?

select sum(t.unit * t.price) from transactions

Search Discussions

  • Santha Kumar at May 13, 2013 at 7:15 am
    Hi Marcel,



    We are facing an issue with respect to "*unsupported column type
    array<string> " . *

    We are able to run the same query in Hive, but in impala we are facing this
    error.

    Ignored the column which contains the *array<string> *data type and did a
    select with the remaining columns from the same table, but still getting
    the same error.

    Not able to access that tables itself.



    Please verify and help us out in resolving the issue.



    *Regards,*

    Santhakumar
    On Tuesday, April 30, 2013 3:48:12 AM UTC+5:30, Marcel Kornacker wrote:

    Sorry, at this point complex data types are not supported in Impala.
    Support for this will be added at a later date.

    Marcel

    On Mon, Apr 29, 2013 at 3:16 PM, <bewan...@gmail.com <javascript:>>
    wrote:
    If I have a set of objects of Transaction packed into Parquet format, how
    can I sum the total sales of all transactions in Impala?

    class LineItem {
    val unit: Int
    val price: Double
    }

    class Transaction {
    val lineItems: Array[LineItem]
    }

    will this query work?

    select sum(t.unit * t.price) from transactions
  • Henry Robinson at May 13, 2013 at 4:21 pm
    Hi Santhakumar -

    Impala does not currently support tables with compound column types, such
    as maps and arrays. Support for these types will be added in a future
    release. Until then, I'm afraid you will need to exclude the array<string>
    column from your table.

    Best,
    Henry

    On 13 May 2013 00:15, Santha Kumar wrote:

    Hi Marcel,******

    ** **

    We are facing an issue with respect to "*unsupported column type
    array<string> " . *

    We are able to run the same query in Hive, but in impala we are facing
    this error. ****

    Ignored the column which contains the *array<string> *data type and did a
    select with the remaining columns from the same table, but still getting
    the same error. ****

    Not able to access that tables itself. ****

    ** **

    Please verify and help us out in resolving the issue.****

    ** **

    *Regards,*****

    Santhakumar****
    On Tuesday, April 30, 2013 3:48:12 AM UTC+5:30, Marcel Kornacker wrote:

    Sorry, at this point complex data types are not supported in Impala.
    Support for this will be added at a later date.

    Marcel
    On Mon, Apr 29, 2013 at 3:16 PM, wrote:
    If I have a set of objects of Transaction packed into Parquet format, how
    can I sum the total sales of all transactions in Impala?

    class LineItem {
    val unit: Int
    val price: Double
    }

    class Transaction {
    val lineItems: Array[LineItem]
    }

    will this query work?

    select sum(t.unit * t.price) from transactions

    --
    Henry Robinson
    Software Engineer
    Cloudera
    415-994-6679
  • Andrew Stevenson at Aug 27, 2013 at 7:03 am
    Do you have a timeframe? Using impala to query nested avro ingested by Flume would be great. At present i have to use HIVE or PIG to flatten and convert to parquet for query performance. If i could write a flume avro event to parquet and query the complex types in impala it really would be a killer app for low latency analytics.

    To unsubscribe from this group and stop receiving emails from it, send an email to impala-user+unsubscribe@cloudera.org.
  • Marcel Kornacker at Aug 27, 2013 at 2:29 pm

    On Tue, Aug 27, 2013 at 9:03 AM, Andrew Stevenson wrote:
    Do you have a timeframe? Using impala to query nested avro ingested by Flume would be great. At present i have to use HIVE or PIG to flatten and convert to parquet for query performance. If i could write a flume avro event to parquet and query the complex types in impala it really would be a killer app for low latency analytics.
    We are planning on adding support for nested types in Impala 2.0,
    which we expect to go out in Q1 of 2014.

    To unsubscribe from this group and stop receiving emails from it, send an email to impala-user+unsubscribe@cloudera.org.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupimpala-user @
categorieshadoop
postedApr 29, '13 at 10:16p
activeAug 27, '13 at 2:29p
posts5
users5
websitecloudera.com
irc#hadoop

People

Translate

site design / logo © 2021 Grokbase