locked
Slow performance when opening views and scrolling through data RRS feed

  • Question

  • Hi,

     

    We're experiencing some slow and odd performance issues on a few PCs when opening views in ProClarity v6.2 Desktop Professional.

     

    It can take a minute or 2 for the data to appear in the windows pane on the right hand side. This could be deemed acceptable however the annoyance is that when you then start to scroll through the data it doesn't do it smoothly it hangs for around 10 seconds when you release the scroll bar. The pane becomes blank until the data reappears.

     

    I can see in task manager the CPU level shoots up to over 50 percent when the data is being returned and the memory usage level also rises to a high level, around 600 MB. Once the data is returned the CPU returns to normal level but when scrolling it rises to over 50 percent again. 

     

    Strangely this will also happen if there is some other application window on top of the right hand pane. For instance if I have task manager open and it is covering part of the data, I can move the task manager window around and it will turn the right hand pane completely white. Move task manager to the other side of the screen so I can see the right hand pane again and it will hang for 10 seconds, CPU rise while it refreshes.

     

    Has anyone else experienced the above or have any possible solutions? The machines ProClarity are running on are relatively high-spec: Intel dual core, 2 Gb RAM, Windows XP SP3

     

    Any thoughts greatly appreciated

    Wednesday, July 9, 2008 4:40 PM

Answers

  • I think you might be seeing behavior that's a result of the grid architecture.  The ProClarity Pro clients page data for large grid views, so the grid is essentially re-populated with data when the user scrolls out of the immediately visible cell range.  The was especially noticeable in AS 2000 with calculated members in the view.  Since most calculations were done on the client in AS 2000, when the user scrolls in the grid, the news cells are actually being calculated and then displayed.  Depending on how many calculated members there are, and the complexity and size of the data being handled, this could certainly take ten seconds or more.

     

    An OLAP query that takes two minutes is probably pulling a large dataset.  It also sounds like the users are viewing that information in a grid, which means they are probably trying to see a fairly large, tabular report.  You might consider what the users are really after and see if it's possible to start the view at a higher level, and then use drill down, crossdrill, decomposition tree, etc to get the answers they're really after.  That would narrow down the dataset enough to likely avoid this issue.  If they really do need a report to get the answers they want, you might consider setting them up on Reporting Services or maybe some Excel pivot tables, both of which are designed to handle this kind of dataset much more efficiently than an analysis tool like ProClarity is able to.

     

    Thursday, July 10, 2008 10:29 PM

All replies

  • Hello D,

     

    Sorry to hear you're experiencing problems with performance.  It sounds as if we can rule out a lack of resources, which is typically what causes these symtpoms as you well know.  If you're running 6.2 I highly recommend upgrading to the latest 6.2 hotfix version (6.2.2215).  If you still experience problems, can you narrow it down to one query in particular?  Or does the issue occur regardless of what query is being executed with the Professional client?  And does this happen when you're connected via TCP/IP to the cube server (File / Open Cube) or when you're working with PAS briefing books?

     

    Another thing you might check is to ensure your anti-virus software is not interfering.  Are you able to disable the anti-virus and test the performance? 

     

    And are you using AS 2000 cubes or SSAS 2005 cubes?  In either scenraio, you'll want to ensure that your AS machine is up to date with the latest service pack as well as the client components on the client machine. 

     

    If you have anymore information to pass along I'd be happy to see what I can suggest.

     

    Kind Regards,

    Amanda

    Wednesday, July 9, 2008 8:28 PM
  • Thanks for the response. Where can I get hold of the hotfix you mention? The link to download software on the old proclarity support site doesn't appear to work and I can't find it in the microsoft download center.

     

    BTW we are using AS 2000 cubes (patched to SP4 as is AS on the client machine). And we connect over tcp/ip to the cube catalogue

     

    Thanks again

     

     

     

     

     

    Thursday, July 10, 2008 10:16 AM
  • I think you might be seeing behavior that's a result of the grid architecture.  The ProClarity Pro clients page data for large grid views, so the grid is essentially re-populated with data when the user scrolls out of the immediately visible cell range.  The was especially noticeable in AS 2000 with calculated members in the view.  Since most calculations were done on the client in AS 2000, when the user scrolls in the grid, the news cells are actually being calculated and then displayed.  Depending on how many calculated members there are, and the complexity and size of the data being handled, this could certainly take ten seconds or more.

     

    An OLAP query that takes two minutes is probably pulling a large dataset.  It also sounds like the users are viewing that information in a grid, which means they are probably trying to see a fairly large, tabular report.  You might consider what the users are really after and see if it's possible to start the view at a higher level, and then use drill down, crossdrill, decomposition tree, etc to get the answers they're really after.  That would narrow down the dataset enough to likely avoid this issue.  If they really do need a report to get the answers they want, you might consider setting them up on Reporting Services or maybe some Excel pivot tables, both of which are designed to handle this kind of dataset much more efficiently than an analysis tool like ProClarity is able to.

     

    Thursday, July 10, 2008 10:29 PM