XenDesktop Bandwidth: The Complete Set
Part 1 – The Prologue: Methodology and Infrastructure
Part 2 – By The Numbers: Take the time to optimize
Part 3 – Bringing It All Together: Daily User Average and General Recommendations
Part 4 – What about XenApp?
Part 5 – Do It Yourself: Starter Kit
Part 7 – Who needs a part 6 when we’re discussing XenDesktop 7
Part 4: What about XenApp?
In my last post, I discussed some general recommendations for a WAN deployment of XenDesktop 5.6, but what about XenApp?
In this post, I’ll be comparing bandwidth requirements between XenApp and XenDesktop and show how XenApp can have a part to play when planning for a WAN. Tests were performed using XenDesktop 5.6 FP1 and XenApp 6.5 FP1.
XenApp Hosted Shared Desktop or XenDesktop
I started out by repeating some of the previously discussed tests performed on Windows 7, except this time on Server 2008 R2. Once again, there were three configurations; listed below.
- OOBE: A default “out of the box” configuration
- Desktop Style: A configuration as close to that of the Default setup previously tested for XenDesktop on Windows 7
- Max Optimized: A configuration as close to that of the “Pushing The Limit” setup previously tested for XenDesktop on Windows 7
Note: There is no guarantee that the 2008 R2 configurations are identical to the Windows 7 ones because the operating systems are different.
These configurations were tested with the same tests outlined in post 2. The results are summarized in the charts below and as can be seen once again, bandwidth can be considerably reduced with optimizations and policies. Interestingly, the desktop style configuration did not add significant bandwidth compared to the default configuration although no policies were set for either of the configurations. When disabling all visual settings and reducing the frames per second (FPS) the bandwidth dropped significantly.
In comparing XenDesktop and XenApp desktops, I decided to focus on the most restrictive configuration as that would be of interest when bandwidth was the biggest issue. In doing so I found some mixed results. On a 1.536Mbps network, XenApp exhibited bandwidth as low as 14kbps for Microsoft Excel and Word, lower than the approximately 20kbps seen for XenDesktop. Note that the user experience in both cases were similar for XenApp and XenDesktop. When looking at Adobe and PowerPoint however, bandwidth was slightly higher proving once again, that you should test, test, test prior to production. Where XenApp can really play a role in a WAN deployment is when only applications are needed and is discussed below.
Publish What You Need
What if users do not require a full desktop? In some situations, a single application is required at a remote location with extreme network conditions. If so, simply provide what is needed. By accessing an application directly, there is no effect in terms of bandwidth when minimizing or dragging the window. This reduces the amount of screen updates required which can lead to lower bandwidth utilization. Furthermore, now that the entire desktop is not being delivered, all visual effects can be disabled without effecting user experience (think of the Max Optimized configuration discussed in the second post in the series). To illustrate this consider the images below.
The first image is a hosted shared desktop delivered through XenApp 6.5 FP1. To conserve bandwidth the Max Optimized configuration was used, as can be seen by the lack of desktop background and the classic taskbar. As mentioned previously in this series, users expect a rich experience and this classic look should be considered in more extreme network conditions. A user might logon to this desktop to edit documents on Microsoft Word and Excel and may be disappointed by this basic appearance. Since nobody wants disappointed users, I present you with an alternative.
In this second image, the same applications were opened as on-demand apps. Once again the Max Optimized configuration was used, but this time only the applications were launched and the user can use his or her local desktop. This can help reduce bandwidth further by eliminating many of the other tasks the user might perform in the virtualized infrastructure as they are no longer available. The on-demand apps FlexCast model is not always practical depending on the situation, but it’s a great option for users with limited network conditions and a small number of active applications.
As always, this scenario was tested to confirm the bandwidth usage. See the results summarized below for a 1.536Mbps network. Bandwidth differences were minimal between the published desktops and on-demand apps (MS Word was less than a 1kbps difference!) as the same workloads were tested for comparison.
The results are also very similar because the Login VSI workloads did not have interaction with the desktop. The user, however, as mentioned above, would have a different experience in the two scenarios. In the case of the published desktop, the user would see a bare desktop. In the case of the Seamless applications, the user would see his or her personal desktop with whatever customizations are in effect. This is all achieved at the same bandwidth requirement for the application without the additional bandwidth of interacting with the desktop.
Now you might be thinking, “wait my application is not compatible with XenApp for reason XYZ”. In that case I present you with two additional options. You can either leverage Citrix AppDNA application management software to help with testing and remediation of compatibility issues, or you can use the VM Hosted Apps feature of XenDesktop.
Next Time
In the next post, I’ll be sharing specific policy and registry settings to help you get started with optimizing your own WAN environment.
Thanks For Reading,
Amit Ben-Chanoch
Worldwide Consulting
Desktop & Apps Team
Project Accelerator
Virtual Desktop Handbook
Conde Malagueta 