My Take on Microsoft’s Announcement to Retire Azure RemoteApp

On Friday, August 12, 2016, Microsoft made the announcement that they are planning to retire Azure RemoteApp (ARA). According to the article, Microsoft and Citrix are jointly developing a new cloud-based solution for delivering Windows apps. This Azure RemoteApp 2.0 service was referred to as XenApp “express”. Only minutes after the Microsoft announcement, Citrix posted a related article on their blog. Now, after the dust has slowly settled, I wanted to share my thoughts on the announcement.

But first things first, here is a list of the important articles covering this topic.

The most important statements in the article published by the Microsoft RDS Team are that Citrix XenApp “express” (or ARA 2.0) is intended to combine the simplicity of application remoting and the scalability of Azure with the security, management, and performance benefits of XenApp, to deliver Windows applications to any employee on any device. New purchases of the current version of Azure RemoteApp will end as of October 1, 2016. Microsoft will continue to support existing Azure RemoteApp customers on the service through August 31, 2017. Citrix added in their article that “Microsoft recently announced its intent to wind down the Microsoft Azure RemoteApp service and assist Citrix in its effort to develop the next generation of the service.” This means that ARA 2.0 is not ready, yet. We’ll have to wait a little longer until we see a beta or a technology preview.

Okay, what’s my take on this? Let’s start at the very beginning. Since ARA was released, I’ve tested and used it many times, including an ARA benchmarking project initiated by the RDS product group in 2015. At the end of this project, fellow MVP Kristin Griffin and I wrote a Microsoft-internal whitepaper (with Freek Berson and Ruben Spruijt as reviewers) that summarized our results and suggested necessary changes. In several follow-up calls we tried to explain our change requests, but most of our recommendations were ignored. Without blaming anybody, it was the typical IT pro versus developer dilemma. On “our” side: infrastructure experts and IT architects with 10+ years’ experience in RDS and Citrix environments. On “their” side: developers and program managers who had never seen a large RDS production environment before. We simply didn’t speak the same language…

We had the impression that the RDS product group was struggling with understanding use cases and product requirement we were asking for. Even after creating an “RDS product group-friendly” version of our whitepaper, it was never published. Still, the RDS PG was always very open-minded and they used our input to improve certain aspects of ARA over time – unfortunately, this was not enough. But they gave us permission to share some of the project results with the public. This allowed me to (co-)present the good and the not so good things about ARA at a number of conferences and industry envents. And some RDS experts started helping customers with their ARA projects, finding out the hard way that ARA doesn’t scale as expected.

Now, what is the problem with ARA 1.0? Despite the great idea, ARA has several architectural shortcomings and missing features that negatively influence scalability, load balancing, user assignment, administrator experience and application management. It’s a great solution for application testing, demo environments, proof-of-concept installations and for homogeneous groups with less than 500 users and only small application sets. But this is clearly not enough for Microsoft’s ambitions. Their goal was to establish an enterprise-ready Azure hosting service for conventional Windows applications that is suitable for millions of users. From this perspective, Microsoft retiring ARA 1.0 and joining forces with Citrix to build ARA 2.0 makes lots of sense. Citrix XenApp/XenDesktop has all the necessary product features that Microsoft Azure is missing, like load balancing suited for Windows apps, (virtual) network components (NetScaler), scalability (FlexCast), superior image deployment methods (MCS and PVS), integrated disk layering (Personal vDisk), application catalogs and sophisticated user assignments.

In addition, Citrix’ new CEO Kirill Tatarinov was Corporate Vice President of the Microsoft Business Solutions Division until October 2015. So ARA 2.0 is a deal among friends and former colleagues. But there is still a risk. I agree with those people who think that the announcement to retire ARA is not good for the Azure reputation and may keep customers away from the upcoming new ARA service. It’s an important lesson for the Azure team to learn: make sure a new service is well-designed and take responsibility to deliver it in a consistent way to your enterprise customers. Communication and timing of announcements are critical factors, as we are seeing now.

So let’s hope that Microsoft and Citrix can get the new ARA service off the ground together. As weird as it may sound, none of them has the experience of planning and building large, global RDS environments. This has always been done by partners. But there is enough experience in the field to build large environments for publishing Windows applications with Microsoft RDSH and Citrix XenApp. So chances to succeed are much better now – and I’m looking forward to learn about roadmap and delivery dates. I (silently) wished that this time Citrix and Microsoft would be listening more carefully to architecture recommendations provided by experts in their respective technical communities. And I hope that Microsoft Ignite will shed some light on the future of ARA.

It is my strong belief that mid to long term native Web apps and mobile apps will replace most conventional Windows apps, across all market sectors and enterprise sizes. But until then, Windows applications hosted on Azure will be a viable option for quite a while – with ARA 2.0 and GPU accelerated Azure VMs (N-series) as important pieces of the puzzle. For me as a remoting expert and enterprise architect it’s exciting times.