Cube processing 30x slower than other cubes of same kind

  • i have created about 20 cubes each in their own adb, and each against a diff source rdb (but with like schemas), and one of them takes 30x longer to build than others of similar data volume.  has anyone experienced this before? if so, do you have any suggestions as to how to solve?

    i've stripped away most dimensions and the rebuild still takes way longer than i'd expect.  the slowness happens when it is chugging through the rows.  it processes at about 2000 rows / second, where as other cubes are ripping through at about 60000 rows / second.  all of these cubes use the same fact view (same query and columns against the same table structures), just against different db's.

    analysis server version: 8.00.382, SQL server version: 2000 - 8.00.760 std edition Win NT 5.2 (Build 3790)

    any help is appreciated.

    thanks, Paul.

  • Take the query it's using to read the data (double click the little sql box) and put it into query analyzer to see the execution plan. There must be something that needs an index.

  • thanks for the feedback.  i agree, but i'm not sure which little sql box you are referring to.  is this found somewhere in the Analysis Manager?  does it give me the sql behind the way that the fact table and dimensions join together to build the cube?

  • When you process the cube, a window will show all the processes. Don't close that window after the process finish. You can double click on the queries in this window and copy to SQL Analyzer and run them.

    Jolene

  • And change your server properties to always log cube builds -> right click on the server, select properties and then set the log file to be somewhere you'll remember it.

    Steve.

Viewing 5 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply