5 Replies Latest reply: May 20, 2016 9:40 AM by Sangram Reddy RSS

    Qliksense architecture

    Ashvin Ramiah

      Hi All,

      Would you please shed some lights in the following :

      I have one Customer who has the following requirement :

       

       

      Number of Dashboards

      Number of Sheets per Dashboard

      Estimated number of qvds

      Number of users

      Company 1

      5

      7

      21

      10

      Company 2

      5

      7

      21

      15

      Company 3

      6

      5

      18

      10

      Company 4

      5

      5

      15

      10

      Company 5

      5

      4

      15

      10

      Company 6

      7

      5

      21

      10

       

      Average qvd per dashboard = 3

      Total Number of qvd = 99

       

       

      They have 8 companies and 7 users per/company and their ERP are centralized.

      All users will be connected and they need high-availability/ no downtime  on both Qlik and the infrastructure that we have to recommend them.

      It should be a like 50 concurrent users will be accessing Qliksense Ent, through a private WAN. In future it will grow to more than a 100 users!


      The following are some queries...

      1. How many tokens they need, if they have 2 Server( nodes ) Assuming there will be 50 users.

      2. Can someone propose the best solution architecture required , in terms of server capacity etc.

      3. How to handle load balancing

      4. How to handle fail over of one server at any given point of time?




      Any other suggestion would be most welcome.


      Regards,





        • Re: Qliksense architecture
          srinivas jonnakuti

          1. There is required Pre requests for Above scenario. Is it all above companies under Single domain[Windows AD, Network]. According to above table need 65 tokens

          2.

          FileSizeMultiplier is the RAM Usage of Single Qliksense Application. Ex: 1GB of Qliksense Application will take 4 GB of RAM.

          userRAMratio is the RAM Usage of additional Users after the first User.

           

          userRAMratio: range between 1%–10%

          FileSizeMultiplier: range between 2–10

          CompressionRatio: range between 20%–90%

           

          3.through BinarySync to handle load balancing

          4. By checking services running or not, through monitor apps

          • Re: Qliksense architecture
            Sangram Reddy

            HI Ashvin,

             

            1) How many tokens they need, if they have 2 Server( nodes ) Assuming there will be 50 users?

             

            As you must be aware of, there are 2 types of tokens

            a) User access token

            b) login access token

            According to your explanations, your users look like they intensely use the analytics platform and I would recommend you to go with 60 licenses to begin with and assign a User access token initially. As you have apps prebuilt to govern the licensees and operations of users, you will be able to move the infrequent users to a Login access pass accordingly.

            Based on my experience 100 Licenses should provide access to 150 users (+ or - 10 users) roughly.

             

            2) Can someone propose the best solution architecture required , in terms of server capacity etc?


            I would propose a 32GB ram and a 16 core.


            3) How to handle load balancing.


            You can go with a cluster nodes to balance the load:

            Qlik Sense Multi-node Setup Guide


            4) How to handle fail over of one server at any given point of time?

            For this, I have a solution i have developed for my company. Any error will be logged in the log files for sure and If you can read the log files on a very frequent basis using a python script, you will be able to automate the process of sending you alerts to your team regarding any failure in Qliksense like the task failures, services going down, new users trying to login in etc......

            Check this for an example I have posted : Automatic Alerts(Email)

             

            Hope this helps.

             

            Thanks,

            Sangram.