User Personalization Layer – First Impressions

by Benjamin Crill, CTA

NEW LTSR!!!!!

Citrix gave customers a year-end present by releasing the newest Citrix Virtual Apps and Desktops (CVAD) long term service release (LTSR). Enterprises have been clamoring for this update to get so many of the updated features that have been released in the current release line since 7.15. It’s been two and a half years and Citrix has released many solid feature updates. For a large enterprise, like where I work now, being on current release is not feasible; so, a new LTSR is a welcome present.

User Personalization Layer

One of the new features that is truly a welcome surprise, as I don’t even recall it being hinted at, was the addition of user personalization layer (UPL). It is the replacement for Personal vDisk (PvD), which has been deprecated, for having a non-persistent machine image while allowing user controlled and installed applications. Think of a world where as a Citrix admin, you can have your stable of virtual desktops that are centrally managed AND you can allow users to install applications. With a significant amount of application development being provided by offshore third parties, companies often look at VDI as a way to secure the access of the contractors and keep data within their control. The problem comes in that each development team, and really each developer themselves, uses a different toolset. Depending on what project they work on, and what part of the project they are responsible for, the toolsets could be totally different. What this ultimately results in is a lot of persistent desktops. One can’t get too upset by using persistent desktops because they still get the other benefits of Citrix, but it does make it harder to support. It is messy, but it gets the job done.

While on the consulting side, I had several customers try to leverage PvD to solve this persistent virtual desktop problem. PvD had the concept right, but where the rubber meets the road, it didn’t fit. The indexing process seemed to be fragile. Without a proper indexing, the PvD wouldn’t mount and the user just had a regular non-persistent desktop. Also, the way that PvD worked tied a user to a single desktop. Yes, there were ways to move the PvD, but it wasn’t a process that scaled well. Thankfully, DR was out of scope for PvD in the implementations I had.

So how is UPL different? This technology leverages technology from Citrix App Layering to provide a way for users to customize their experience as well as install applications. The advantage of UPL is that it does not require the full App Layering infrastructure. More importantly, the admin does not have to refactor their image into App Layering to use the feature. All that is required is the brokers and VDAs be at 1912 LTSR, you have a file share, and then you set two policies.


I was amazed that it just simply worked. Admittedly, I didn’t even research it, I just configured and tested. With just a base Win10 1909 build and 1912 VDA brokered from Citrix cloud, I had an MCS provisioned Win10 with UPL in minutes thanks to being able to continue using my existing build automation.

Once I logged in, I saw the directory structure being built. Under the share specified, it creates a Users folder. From there is a domain_username folder, and under that is an A_OK folder. Within that A_OK folder is the user VHD as well as log files. Obviously, the top file is the user VHD, and then below are diagnostic logs that show which systems the user has logged into as well as logoff processing. The VHD gets mounted in guest at logon so a user is no longer tied to a single VM.


In my testing I was successfully able to install common productivity apps and have had no problems. I have been quite impressed with how the applications perform as well. One really wouldn’t know that the applications aren’t installed natively.

Going Forward–Conclusions

This is an exciting new feature that many will look to leverage. There has already been some considerable chatter within the EUC forums about it. If it works as advertised it will certainly help the management of virtual desktops. It will certainly not eliminate persistent desktops as there is always going to be a stable of applications that don’t work well dynamically attached. Keeping expectations in line for admins is important. Testing will be key to see how it fairs in more diverse scenarios. A personal lab can only test so much, and the real world has many interactions and integrations that can provide unexpected behaviors.

Another major factor is scalability. This is where mileage will vary considerably. Performance will depend on many factors. Two major factors will be the rate of change (how many and how often users are installing applications) and the performance of the file share hosting the user VHDs. It all boils down to I/O and sizing the file share to provide the necessary performance.

Manageability is also going to need to improve to make it truly scalable. The simplicity of it is certainly a benefit and a drawback. Currently, the only option is to put a share. It would certainly be nice to have more granular capability, leveraging variables, to place the users’ VHD. Also, there is only the option to specify the size making it look vary static. Having the ability to set an initial size and let it grow would be advantageous so that admins don’t have to use a size that will cover all. Granted, multiple policies could be used, but it still means that all users will be driven by the heaviest user.

Overall, it is a step in the right direction. At a minimum, it achieves simplicity in the solution which was lacking before. Adoption will be the true reflection of how well the problem has been solved, but it is encouraging to see. I truly hope the Citrix continues to develop and improve upon this feature to make it enterprise ready. Now go and give it a shot for yourself!

Benjamin Crill
@benjamincrill
www.icrill.com

Leave a Reply