FAQ
Hi

1)Is there any paper on how impala works?. Are those Hive queries converted
into
HBase scans internally? Or does impala have some other mechanism to retrieve
the hbase data?. Any internal paper describing the architecture in detail?

2)Has someone tried making impala work with mondrian? From googling,
i understand there were some pending issues with the impala driver from
the cloudera side. Can you kindly tell me whether these issues have been
fixed? What is the best way to go about integrating mondrian with impala?
http://jira.pentaho.com/browse/MONDRIAN-1424

Thanks
Karthik

Search Discussions

  • Justin Erickson at May 27, 2013 at 8:07 pm

    On Mon, May 27, 2013 at 1:02 PM, Marcel Kornacker wrote:
    On Mon, May 27, 2013 at 4:42 AM, Karthik wrote:
    Hi

    1)Is there any paper on how impala works?. Are those Hive queries converted
    into
    HBase scans internally? Or does impala have some other mechanism to retrieve
    the hbase data?. Any internal paper describing the architecture in
    detail?

    Karthik, there is no published paper on the internal architecture of
    Impala at the moment - we feel that's somewhat premature, considering
    that things are still evolving at a fast pace, but we are planning on
    providing an in-depth discussions of the internals once things have
    settled down a bit.

    Regarding your specific questions: a query against an hbase table ends
    up executing scans against that table using the standard hbase client
    library.

    As Marcel says, when Impala is querying against HBase data it will leverage
    the HBase region server and APIs to perform scans of the HBase tables. When
    possible it will push down predicates such as primary key predicates to
    minimize the scan ranges. Note that Impala can query HBase, but HBase is
    not necessary for non-HBase data.

    We don't yet have papers on Impala but you can find more material on Impala
    in the following links:

        - Impala 1.0 GA blog:
        http://blog.cloudera.com/blog/2013/05/cloudera-impala-1-0-its-here-its-real-its-already-the-standard-for-sql-on-hadoop/
        - Impala E-Learning Course:
        http://training.cloudera.com/elearning/impala/
        - Rich Report Impala overview:
        http://hpcradio.blogspot.com/2013/05/technical-overview-of-cloudera-impala.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+SunRadioHpcPodcast+(The+Rich+Report+HPC+Podcast)
        - Tech dive into Impala webinar:
        http://www.cloudera.com/content/cloudera/en/resources/library/recordedwebinar/cloudera-impala-a-modern-sql-engine-for-hadoop-video-recording.html
        - Impala/Cloudera Enterprise RTQ site: http://www.cloudera.com/impala
        - Impala Docs and FAQ:
        http://www.cloudera.com/content/support/en/documentation/cloudera-impala/cloudera-impala-documentation-v1-latest.html


    Regarding the question on Pentaho and Impala:
    2)Has someone tried making impala work with mondrian? From googling,
    i understand there were some pending issues with the impala driver from
    the cloudera side. Can you kindly tell me whether these issues have been
    fixed? What is the best way to go about integrating mondrian with impala?
    http://jira.pentaho.com/browse/MONDRIAN-1424


    We've been working closely with Pentaho. We have a tested out a solution
    but need to fix IMPALA-85 before the Pentaho on Impala is ready for
    non-beta production usage. We're hoping to fix this integration issue this
    summer.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupimpala-user @
categorieshadoop
postedMay 27, '13 at 11:42a
activeMay 27, '13 at 8:07p
posts2
users2
websitecloudera.com
irc#hadoop

2 users in discussion

Justin Erickson: 1 post Karthik: 1 post

People

Translate

site design / logo © 2022 Grokbase