+ Reply to Thread
Results 1 to 2 of 2
  1. #1
    Junior Member
    Join Date
    Mar 2011

    Optimizing Infobright EE

    Hi Infobright Community,
    i have a problem, maybe someone could help me. I am testing Infobright CE, Infobright EE and InfiniDB CE. In my test scenario Infobright CE is slow and Infobright EE is as fast as InfiniDB CE. InfiniDB is just using one server node of three. If i am using all nodes the System is much faster. My test system is based on Red Hat 5.8. The server has 32 GB RAM and four dual core processors. Infobright EE take advantage of 24 GB RAM.I installed Infobright as described in the instructions. Infobright EE is using port 5029. Its the only service on this port. The table structure looks like:

    CREATE TABLE facts_flat
    area VARCHAR(10)
    , region VARCHAR(7)
    , article VARCHAR(10)
    , articlegroup VARCHAR(15)
    , subgroup VARCHAR(4)
    , customer VARCHAR(15)
    , customertype VARCHAR(8)
    , customergroup VARCHAR(3)
    , store VARCHAR(10)
    , country VARCHAR(13)
    , postal VARCHAR(15)
    , trademark VARCHAR(6)
    , date VARCHAR(15)
    , w_day VARCHAR(10)
    , year BIGINT
    , month BIGINT
    , day BIGINT
    , turnover BIGINT
    , amount BIGINT
    , revenue BIGINT
    , discount BIGINT

    I make performancetests with native queries and Pentaho Mondrian.
    The queries look like this:

    Select area, article, sum(turnover)
    From facts_flat
    Where area = 'Germany'
    Group by area, article;

    What can I do, to optimize Infobright EE?

    Thank you

  2. #2
    It sounds like you have a great environment for IEE to perform well in. m Some of the things I would look into trying to do is determine which columns you can use LOOKUP with. Depending upon the distribution of unique data, a LOOKUP column could hugely increase query performance. For instance, area = 'Germany' , this sounds like it could be a great candidate for a 'LOOKUP'


    More information on optimizing queries can be found here


    We have a tendency to read the bh.err file when we start to benchmark and run performance testing because we like to understand how the query execution will break down and this bh.err will let us know how the execution path is performing.


    This is some great starting points. If there is a direct issue, creating a ticket in the helpdesk would be your best course of action.
    Last edited by craigtrombly; January 27th, 2014 at 04:16 PM. Reason: spelling



Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts