Citrix Virtual Apps and Desktops – CPU Utilization


With the 1912 release of Citrix Virtual Apps and Desktops 1912, we saw a major improvement in overall bandwidth utilization.  Some of the tests showed an 80% reduction in bandwidth utilization.  But how does this drastic decrease in bandwidth utilization impact CPU utilization?  Let’s find out.

Test Details

The test scenarios are as follows:

  • Workloads: LoginVSI v4.1.39
    • Task Worker
    • Knowledge Worker
    • Power Worker
  • Operating System:
    • Windows 10 (1903)
    • Windows 2016
  • Optimizations: Citrix Optimization Tool-1903 template
  • Policies: (policy settings defined at the end)
    • Baseline
    • Bandwidth
    • User Experience
  • Test Duration: 60 minutes
  • Test Iterations: 3 cycles
  • Metric Sample Period: 5 seconds

CPU Utilization Results

Windows 10

Let’s look at the CPU utilization for a 60 minute LoginVSI test within Citrix Virtual Apps and Desktops for a single user on Windows 10.

The first set of results are using the baseline policy.

ReleaseTask (Baseline)Knowledge (Baseline)Power (Baseline)
7.154.04%5.22%5.88%
19123.4%4.44%5.07%
Windows 10 – Average CPU for 1 User (Baseline Policy)

The second set of results uses the bandwidth savings policy.

ReleaseTask (Bandwidth)Knowledge (Bandwidth)Power (Bandwidth)
7.154.23%5.09%6.09%
19123.25%4.47%5.22%
Windows 10 – Average CPU for 1 User (Bandwidth Savings Policy)

And the third set of results uses the user experience policy.

ReleaseTask (Experience)Knowledge (Experience)Power (Experience)
7.154.34%5.15%6.078%
19123.6%4.44%4.95%
Windows 10 – Average CPU for 1 User (User Experience Policy)

Windows 2016

In the next set of data, we look at a 60 minute LoginVSI test within Citrix Virtual Apps and Desktops with 10 users on Windows 2016

The first set of results uses the baseline policy.

ReleaseTask (Baseline)Knowledge (Baseline)Power (Baseline)
7.1516.8%30.59%38.27%
191215.65%24.75%29.04%
Windows 2016- Average CPU for 10 Users (Baseline Policy)
ReleaseTask (Bandwidth)Knowledge (Bandwidth)Power (Bandwidth)
7.1520.79%35.68%34.47%
191215.85%24.81%31.38%
Windows 2016- Average CPU for 10 Users (Bandwidth Savings Policy)
ReleaseTask (Experience)Knowledge (Experience)Power (Experience)
7.1516.93%30.75%35.36%
191215.71%26.10%29.9%
Windows 2016- Average CPU for 10 Users (User Experience Policy)

If we graph these results, you can quickly see the reduction in overall CPU utilization across all test scenarios. This first graph shows 1 user on Windows 10 1903

The second graph shows 10 users on Windows 2016

Go back and take a look at the bandwidth numbers. When you compare 1912 release to 7.15, 1912 has lower bandwidth and also lower CPU utilization.

Over the years, just as the protocol became smarter about how to reduce bandwidth, the protocol also became much more efficient.

Policy Settings 

PolicyBaselineBandwidthExperience
Audio QualityHighLowHigh
Desktop wallpaperAllowedDisabledAllowed
Dynamic windows previewEnabledProhibitedEnabled
Limit video qualityNot configuredMax 480p/720kbpsNot configured
HDX Adaptive Transport7.15: Off
1912: Preferred
PreferredPreferred
Menu animationAllowedProhibitedAllowed
Preferred color depth for simple graphics24 bits per pixel16 bits per pixel24 bits per pixel
Target frame rate30 fps16 fps30 fps
Target minimum frame rate10 fps8 fps10 fps
Use video codec for compressionUse when preferredFor actively changing regionsUse when preferred
View window contents while draggingAllowedProhibitedAllowed
Visual qualityMediumLowHigh

Daniel

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.