2 Replies Latest reply: Apr 15, 2015 7:43 PM by Sergey Makushinsky RSS

    Lack of performance on QV Server (need advice!)

    Nikita Kokurins

      Hi everyone!

       

      I wonder if someone could suggest me how to increase performance of QV server in my situation.

      We are running QV SBE server with 2.8Ghz 4-core 64GB RAM server.

      I'm now working in project of enriching our analytics with geo component. We're having a KML file with ~600k square blocks, population attribute for each of them and a customer base of ~1 mil people living inside those blocks.

       

      Loading this KML file using QuickMap extension was not successful, but that's clear why: QuickMap extension uses front-end approach and is not meant to work with huge KML files like ours (~330MB). We found a solution for this, I think some of us might be interested in that, cause I didn't find similar solution on community. We took only first (left upper) block's coordinate, calculated apprx coordinates of the center of the block and used them as a layer in map object (this is a standard google map solution for qv, you can find many manuals in community).

      So, having only one coordinate for each block we placed them on the map and used bubbles for visualization, bubble sizes were adapted similar to "block" sizes (thus they are next to each other) basing on the zoom variable. We used macro for bubble sizes based on current zoom. I can share it if someone is interested.

       

      Eventually this solution worked!

      We are now able to see the map with visualized "blocks" and necessary data, BUT there's an issue we came up with. Response time is still quite slow, especially when it's zoomed out. It' s clear why, qv has to process ~600k bubbles in one object and I bet that' s not something expected to be processed in 0.01 seconds...Nevertheless, our users are not satisfied and I'm now thinking of what could help to get rid of this performance issue. In fact, I've found out that increasing server capabilities (especially increasing the number of CPU cores) might solve the issue.

       

      And finally, the question is...can anyone suggest me how to deal with this issue, maybe someone has tried or knows whether increasing CPU cores, speed or RAM might help??

       

      Thank you beforehand.

       

      p.s. by the way, might Qlik Sense be able to process such big KMLs? I know there is some limitation in current version and it's impossible to load big KMLs...will this be fixed in next releases?