Citrix Virtual Apps and Desktops 2003 is the first release after the much anticipated 1912 LTSR. Between the major LTSR releases, there are numerous smaller releases that can directly impact bandwidth consumption. The previous set of tests compared the 1912 LTSR release with the 7.15 LTSR release, showing a huge reduction in bandwidth usage.
The 2003 version was recently released, and that means it is time to see how the bandwidth compares to the earlier release.
Lessons Learned
- Variables: Trying to compare results from release-to-release is extremely challenging due to the numerous changing variables within the environment. For example, in the tests completed 3 months ago, the operating system updated to 1909, new Windows updates were added, Workspace app was updated, the Citrix Virtual Apps and Desktops delivery controller was updated and the VDA was updated. And these are the variables I know about. Comparing this round of tests with a previous round is extremely difficult and not recommended.
- Policies: The different policies (Baseline, Bandwidth and User Experience) have little impact on overall bandwidth usage numbers. You can see this in the graphs from the previous set of tests. There are many reasons for this, but the biggest reason is that the ICA protocol is more intelligent than it was a few years ago. Many of the optimizations for text, video and graphics are automatic based on the current network characteristics. Because these tests are all on a LAN environment with the same latency and packet loss, the Virtual Apps and Desktops policies will have little impact on bandwidth. Based on this, the tests will only focus on a single, baseline policy.
- Glyph Caching: Prior to 2003, glyph caching was enabled. This led to a noticeable reduction in bandwidth usage, especially for task workers. In the 2003 release, the glyph caching optimization was changed to be automatic based on the user’s session performance. Because these tests are executed on a LAN, glyph caching will automatically be disabled, inflating bandwidth numbers for the 2003 release. To maintain consistency between releases, glyph caching was forced “On” for the 2003 tests.
Test Details
The test scenarios are as follows:
- Workloads: LoginVSI v4.1.39
- Task Worker
- Knowledge Worker
- Power Worker
- Operating System:
- Windows 10 (1909)
- Windows 2016
- Optimizations: Citrix Optimization Tool-1909 template
- Policies: Baseline
- Test Duration: 60 minutes
- Test Iterations: 3 cycles
- Metric Sample Period: 5 seconds
Results – Windows 10
First, we look at Windows 10 bandwidth utilization for a 60 minute LoginVSI test using the default policy settings within Citrix Virtual Apps and Desktops.
| Release | Task | Knowledge | Power |
| 1912 | 64 kbps | 205 kbps | 359 kbps |
| 2003 | 63 kbps | 203 kbps | 345 kbps |
Results – Windows 2016
In the next set of tests, we look at Windows 2016 bandwidth utilization for a 60 minute LoginVSI test using the default policy settings within Citrix Virtual Apps and Desktops.
| Release | Task | Knowledge | Power |
| 1912 | 30 kbps | 187 kbps | 373 kbps |
| 2003 | 32 kbps | 182 kbps | 361 kbps |
If we graph these results, you can quickly see the changes in overall bandwidth utilization.
