Performance indicators for JOC Cockpit to display large numbers of objects

Starting Situation

• Performance generally covers the following components:
  • The web service that might access a database to query data.
  • The network that is used to transfer the data from the web service to the browser.
  • The browser that renders results.
• Performance depends on what is measured, what browser is being used and what CPU capabilites are available:
  • Performance measurements are done for the Jobs View, Job Chains View and Orders View. We measure performance for >=200 objects.
  • Browsers provide different performance results, which includes that not one browser product is generally faster than a competitor product but that specific functions will take more or less time. Therefore no data are provided for specific browser products.
  • CPU is crucial as it is used to render objects in the browser.

Performance Measurements

• Measurements are specified for
  • release 1.12.4
  • use with an Intel i7 2.6 GHz CPU with 2 cores. This translates to the fact that 50% of the CPU (1 core) can be used by the client browser process.
  • Gigabit network connection
  • use of Chrome 67 64bit
  • use of ca. 500 MB up to 2 GB main memory by the browser
  • All objects are fully rendered in list view:
    • up to 1000 items in the result list are rendered in a single page. Results are tremendously better if smaller chunks of objects for rendering are used, e.g. displaying only 50 objects per page, see JOC Cockpit profile settings.
    • all details are expanded. Results improve heavily if details are collapsed.
  • The attached PowerShell Script for Windows/Linux Create-PerformanceTestCases.ps1 has been used to create the test objects.

• Display of Job Chains in list view
  • Display a single job chain with details expanded:
    • 200 job nodes: ca. 2s
    • 400 job nodes: ca. 3s
    • 600 job nodes: ca. 6s
    • 1000 job nodes: ca. 6s
  • Display a number of job chains with details expanded:
    • 200 job chains each with 3 job nodes: ca. 20s (i.e. display of 600 job nodes; with details collapsed: ca. 5s)
    • 400 job chains each with 3 job nodes: ca. 50s (i.e. display of 1200 job nodes; with details collapsed: ca. 10s)
    • 600 job chains each with 3 job nodes: ca. 75s (i.e. display of 1800 job nodes; with details collapsed: ca 15s)

• Display of Jobs in list view
  • Display a number of jobs from a single folder:
    • 200 jobs: ca. 2s
    • 400 jobs: ca. 8s
    • 600 jobs: ca. 20s
    • 1000 jobs: ca. 33s

• Display of Orders in list view
  • Display orders for a single job chain with details expanded:
    • 200 orders: ca. 5s
    • 400 orders: ca. 12s
    • 600 orders: ca. 15s
    • 1000 orders: ca. 30s

General Performance Recommendations

• Limit the number of objects per folder
  • Storing some 200 objects, e.g. jobs, in the same folder should be fine for most environments. The exact number depends on your server and client (browser) capabilities.
  • For a larger number of objects use subfolders that can be created for any hierarchy level.
• Check the cores that are available from your CPU.
  • Keep in mind that the browser tab that holds the JOC Cockpit cannot make use of more than 1 core.
  • Therefore if you use 4 cores on a medium sized machine JOC Cockpit can use only 25% of your machines CPU capacity.