I am very fortunate to work for a manager who understands the need for more than one app packaging and delivery product. He understands that each product has its strengths and has suggested nominating a primary solution and a secondary. He is even open to more than two if there is a valid business reason. I have been fortunate to work with pretty much every product on the market for app delivery to a varying degree and am nothing if not opinionated on the subject 🙂
My team have used App-V as our primary app packaging solution for many years. The organization was using App-V long before I even joined the team. App-V is still a great product and has some strengths that are still unmatched by others. For example, I prefer the App-V implementation of running custom scripts more than competing product’s methods and its support by other vendors for product integrations is unmatched. It also happens to be the product I have the most experience with, but we have to face reality. It is not being developed anymore. As my friend says it’s in hospice. It’s being maintained and made comfortable before its inevitable End of Life, don’t expect any major improvements.
Consigned to the loss of App-V, we decided to assess other products to figure out what our new primary solution should be and also what solution could be secondary. The obvious front runner for us on paper was Microsoft MSIX. We are very happy AppVentix customers! Our preference is to continue to use that product for delivering our apps in real-time. AppVentix supports MSIX and MSIX App Attach so we were excited by that prospect.
App-V to MSIX
I have been using MSIX off and on for over 2 years to keep my knowledge and skills up to date. Due to compatibility limitations, I held off on even attempting to introduce it into my work environment. The announcement of some support for services gave me hope that some of the other limitations would be addressed too. When we started to setup a new Citrix Site, I decided this would be a good time to cut over from App-V Scheduler to AppVentix and if we’re introducing AppVentix which can deploy MSIX and MSIX App Attach it seemed logical to revisit MSIX and MSIX App Attach and try them in my work environment.
Video Demo of MSIX
As luck would have it, just as I went on PTO, Microsoft published a PowerShell script to bulk convert App-V 5.1 packages to MSIX. Rather than test with net new packages, I figured this could be a quick win for quickly testing MSIX in our environment. Thank you to Eric at XenAppBlog.com for sharing his updated script that uses MSIX Hero for package signing. It made short work of converting of over 100 App-V packages.
Unfortunately, the results from the first series of testing have not been good. The majority of our applications converted have not worked. If you follow Tim Mangan’s excellent sessions and MSIX Report card you will know he reports about a 40% success rate when just packaging apps. He estimates he will soon be able to get up to 60% by using the Package Support Framework, which Tim has made more digestible with his excellent free product. Even at 60%, this is much lower than the success rate I can achieve with App-V. To see a breakdown of Tim’s findings and compatibility rate, check out the great MSIX Report Cards by Tim.
Some of the issues with our converted apps include things like incorrect application shortcuts, missing launch parameters and incorrect file type associations but these can be fixed with the PSF (A recent fix to the MSIX Packaging Tool may also yield better results for FTAs and Shortcuts). Limitations like requiring elevation for the install of some services doesn’t fit well with our deployment goals. The loss of shell extensions\context menus is a major problem for certain apps. Some apps just straight up fail on launch with errors showing the app is unable to find files or other more generic errors.
While I have confidence MSIX will improve over time. Right now, it’s not where we need it to be to make it our primary app packaging and delivery solution. Really, it’s not suited to be a secondary option either as it won’t work for so many of our apps. MSIX App Attach relies on MSIX packages which means at least for now, there is a low rate of success with that too. We’ll file both away for consideration for the future but with a need to start working on modernizing now we need to find another way forward.
During our Proof of Concept of Numecent Cloudpaging rather than taking apps we successfully deployed with App-V already, we used the time to try apps that we couldn’t deliver with App-V or apps we could but which took a lot of custom scripting. The results were very impressive. I can’t go into too many specifics for obvious reasons, but one app we successfully deployed very quickly is the widely used RightFax product that can be a little tricky due to its Outlook integration and various dependencies. To deploy this with App-V, we had to extract the drivers and most of the dependencies to deploy them separately to our desktops. Using Cloudpaging we could package up and deliver all dependencies together with Cloudpaging AND best of all the print driver works!!
Old Video Demo of Cloudpaging that shows how it handles a print driver
One of the biggest pains in the arse we had when trying our best to react to the COVID work from home surge was dealing with apps like Adobe Acrobat Pro DC that only the selected few users needed and were licensed for. That’s no problem for Cloudpaging. We can deploy to just those users, we don’t need to use App Masking or any other global configuration changes to our desktops. It can be delivered on demand to just those users without any need to update images.
If you are in the same position as us and you want to find an App-V alternative, I hope the chart I created to the right is useful to see how Cloudpaging compared to App-V for us. If you would like a more detailed side by side, feature by feature comparison of Cloudpaging and App-V 5.1, you should check out this WhatMatrix.
As part of our decision process, we would like to choose a product that meets potential future goals like a greater move to the cloud. Cloudpaging integrates into our virtual desktops well on-premises but will that work well if we move our desktops to the cloud? I have been familiar with Cloudpaging for a long time but until we did our PoC, I wasn’t aware of their Content Delivery Network offering. A cloud hosted Cloudpaging delivery!
I’m also very lucky that my manager sees the value in cloud services. We use ControlUp’s cloud offering, we have been going all in on many of Microsoft’s Azure hosted services and it has been very exciting for us. Hosting our applications in the cloud and having them dynamically deliver to our users on their desktops makes a lot of sense. Our workforce is more dispersed than ever. We want to ensure we get away from frequent image updates and ideally, we’d like to lean on cloud services to offload some of the maintenance headaches we encounter with patching and upgrading so many different on-premises services.
Converting App-V 5.1 Apps to Cloudpaging
While I am in no real rush to completely replace App-V, I have been toying with the idea of converting our App-V packages to Cloudpaging with minimal effort. Numecent do have a CLI for their Studio tool which I hope to leverage to automate even further. In the above video, I show how to convert apps manually. When I have it completely automated (which may not be soon as it’s not a priority for my work right now) I will update the video.
As discussed, Cloudpaging provides a level of compatibility we did not get with App-V. It also blows MSIX out of the water for us in terms of first-time success rate. It will speed up our package turnaround time which is important in an environment that can be pretty reactive. The packaging itself is simpler than App-V, MSIX and even MSI packaging meaning it’s easier to train staff to help deliver the apps. It also fits our potential long-term cloud centric goals.
It has become obvious. To help modernize our desktops and better cope with remote work, our primary solution should be Numecent Cloudpaging. Our secondary will remain as App-V, which won’t go into EoL until 2026. I don’t anticipate a need to sequence new apps with App-V but we don’t have to rush to repackage the apps so it can remain a secondary for now. We will keep an eye on the development of MSIX and App Attach with a hope that in future it can become our secondary solution. I hope it will have a higher success rate at least by 2024 to give enough time to convert the App-V apps to MSIX again or repackage those that still won’t convert to MSIX or Cloudpaging. If you haven’t given thought to what you will do after App-V yet, I hope this post is useful to you.